Artificial Intelligence and the Rule of Law

Yadong Cui Editor

Blue Book on AI
and Rule of Law in
the World (2020)
Artificial Intelligence and the Rule of Law
Artificial intelligence is an important driving force of the new round of technological
revolution and industrial change. Artificial intelligence and blockchain have risen to
the level of national policy. While the rapid development of AI brings convenience
to people’s lives, it also poses challenges to global changes and the development
of human society, social governance and legal regulation, such as the leakage of
personal privacy, frequent data security problems, and prejudice and discrimination
caused by algorithms. AI is profoundly changing the world, and it is necessary to
think about how to promote the construction of responsible AI from the perspective
of global governance in the future.
In response to these issues, this series aims to explore and study the new situations
and new issues at the intersection of AI, law and political science all over the world,
to provide a legal basis for the development of AI-related frontier fields and to limit
the development of AI within the framework of law to promote the harmonious
and orderly construction of AI and legal systems. Based on the above purpose,
this series is selected and integrated with multidisciplinary strengths to strengthen
the research on legal, ethical, and social issues related to AI, focusing on laws,
regulations, institutional systems, and ethics for the healthy development of AI. The
authors of this series are all top authoritative experts and scholars in AI, law, political
science, and other disciplines in China, and they have already made considerable
achievements and social reputations in this field of research.
This series has a clear awareness of the problem and proposes specific solu-
tions for the implementation and implementation of AI governance from a multi-
disciplinary and multidisciplinary perspective, covering AI, blockchain, algorithms,
digital currency and other specific segments closely related to law, with cutting-edge,
academic and theoretical value, and filling the gap of similar research in China.
Yadong Cui
Editor

Blue Book on AI and Rule

of Law in the World (2020)
Translated by Cao Yan
Editor
Yadong Cui
Shanghai Law Society
Shanghai, China

ISSN 2731-6238 ISSN 2731-6246 (electronic)

Artificial Intelligence and the Rule of Law
ISBN 978-981-19-3585-5 ISBN 978-981-19-3586-2 (eBook)
https://doi.org/10.1007/978-981-19-3586-2

Jointly published with Shanghai People’s Publishing House

The print edition is not for sale in China (Mainland). Customers from China (Mainland) please order the
print book from: Shanghai People’s Publishing House.
ISBN of the Co-Publisher’s edition: 978-7-208-16580-9

© Shanghai People’s Publishing House 2022

This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publishers, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publishers nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publishers remain neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Editorial Board

Editorial Office Administrator: Cui Yadong.

Deputy Administrators: Ye Qing, Liu Xiaohong, Wang Tao, Shi Weidong.
Editorial Board Members (Sort by last name stroke):
Wang Li, Bai Ning, Feng Benguo, Liu Xueyao, Liu Xianquan, Sun Li,
Yang Hua, Yang Fei, Yang Yin, Zhang Feihu, Zhang Zhijun, Zhang Xiaoling,
Chen Shusen, Lin Zhujing, Zhou Xiao, Zhao Chunxue, Hu Peng, Hao Zuocheng,
Yuan Dibao, Cao Peilei, Dong Liwu, Peng Chengxin.
Editor-in-Chief: Cui Yadong.
Associate Editors: Ye Qing, Liu Xiaohong, Wang Tao, Shi Weidong.
Editorial Office (Sort by last name stroke):
Ma Sijie, Wang Wenjuan, Wang Ranran, Wang Yufan, Wang Jian, Niu Binru,
Xiang Mingjing, Liu Yuyan, Yang Shengyuan, Yang Liuzixian, Li Yan, Wu Tao,
Zhang Bingjie, Zhang Shundong, Chen Jidong, Chen Shusen, Lin Zhujing,
Zhao Yucheng, Zhao Surui, Zhao Tiantian, Hao Zuocheng,
Hao Yuxuan, Hu Rongxin, Qin Kun, Xu Aijing, Gong Sihan, Peng Hui,
Xue Mengbai.

v
Speech at the Release of the Blue Book (2020)

Respected leaders,
Distinguished guests,
Experts and scholars,
Commissioned by Cui Yadong, Director of the Editorial Board, I now release the
Blue Book on AI and Rule of Law in the World (2020) jointly compiled by the
Shanghai Law Society, Yangtze Delta Region Institute of Tsinghua University, East
China University of Political Science and Law, and Shanghai University of Political
Science and Law.
The Blue Book on AI and Rule of Law in the World (2019) was released at the
Rule of Law Forum of the 2019 World AI Conference on August 30, 2019. Upon its
release, the book attracted much attention from all walks of life, and China Central
Television and other media have made detailed reports. On August 31, 2019, at the
closing ceremony of the 2019 World Artificial Intelligence Conference, the confer-
ence released the “On the Construction of Artificial Intelligence Shanghai Hub—
Building a First-class Innovation Ecology Action Plan” (hereinafter referred to as
“AI Shanghai Action Plan”), which clearly listed the Blue Book on AI and Rule
of Law in the World as one of the “seven actions” to build a first-class innovation
ecology of AI in Shanghai.
According to the “AI Shanghai Action Plan”, in December 2019, the Blue Book
Editorial Board started the writing of the Blue Book on AI and Rule of Law in the
World (2020), which took 7 months to complete until the end of June 2020.
The Blue Book (2020) mainly consists of 14 parts: preface, overview of AI and
rule of law development, AI and rule of law development index, 2019 AI and rule
of law development assessment report in China, essence of viewpoint from the Rule
of Law Forum of 2019 World AI Conference, analysis reports (four parts, Part 1:
AI policy and law; Part 2: AI and rule of law theory research; Part 3. AI application
in rule of law; Part 4: AI-related case analysis (China), special reports (six reports,
research on intellectual property protection of AI, research on data security in AI
applications, research on legal issues of face recognition and privacy protection,
research on legal issues related to AI applications in medical field, research on legal

vii
viii Speech at the Release of the Blue Book (2020)

issues related to blockchain technology, empirical research on the cultivation of

interdisciplinary talents in AI and rule of law, the special column on the use of AI in
rule of law response to major epidemics (six articles, “law + AI” in the response to
COVID-19, the application of AI in the fight against COVID-19, the application of
health QR codes in the prevention and control of the epidemic and legal adjustments,
the role of the network direct reporting system in the prevention and control of
the epidemic and legal regulation, the development of the epidemic accountability
information disclosure mechanism, prospects for the construction of judicial big
data to ensure enterprises to resume work and resume epidemic and legal regulation,
the development of the epidemic accountability information disclosure mechanism,
prospects for the construction appendices. The Blue Book (2020) focuses on the
latest development of AI and rule of law in the world from July 2019 to June 2020
by collecting and summarizing the latest information on strategies, policies, laws,
major theoretical studies, practical project applications, and typical judicial cases
in various countries regarding the development of AI and rule of law and provides
readers with an objective, fair, and accurate report through analysis and evaluation,
with charts, illustrations, and cases.
Compared with the Blue Book (2019), the Blue Book (2020) has a more reasonable
structure and richer content and the following new features. First, it has formulated the
AI and Rule of Law Development Index (the first of its kind in the world, hereinafter
referred to as the Index). The index mainly includes the AI Rule of Law Development
Index System, index weights, and assessment methods. Based on this, the paper
evaluates and forms an evaluation report on the development of AI and rule of law in
China in 2019. Second, it focuses on important events in the special column on the
use of AI in rule of law response to major epidemics. Third, it includes more content
by adding AI and rule of law development special reports, expert interviews, and
questionnaires, forming a more reasonable structure. It provides academic reserves
and institutional supply for prospering and leading the study of AI and rule of law.
Due to the limitations of time and resources, there might still be some errors or
mistakes in this book. Therefore, corrections and suggestions from readers will be
welcomed.
The writing and publishing of the Blue Book on AI and Rule of Law in the World
will be done every year. Let us make joint efforts to promote the development of AI
and the rule of law and AI global governance. Thank you very much.

Ye Qing
President of ECUPL
Deputy Administrator of the Editorial Board of the Blue Book
Shanghai Law Society
Yangtze Delta Region Institute of Tsinghua University, Zhejiang
East China University of Political Science and Law
Shanghai University of Political Science and Law
July 2020
Speech at the Release of the Index
and the Evaluation Report

Respected leaders,
Distinguished guests,
Experts and scholars,
Commissioned by Cui Yadong, Director of the Editorial Board, I now release the AI
and Rule of Law Development Index and the Evaluation Report on the Development
of AI and Rule of Law in China (2019).

On the AI and Rule of Law Development Index

In recent years, to prevent and cope with the risks and challenges that AI may bring
and to guarantee the safe, reliable, controllable, and healthy development of AI, AI
and rule of law were born, as was its research. To promote the development of AI
and rule of law systems, create AI and rule of law ecology, prosper and lead AI and
rule of law research, and promote AI global governance, according to the proposal of
authoritative experts in law, science and technology and the actual development of AI
and rule of law, the Blue Book on AI and Rule of Law in the World (2020) formulated
the AI and Rule of Law Development Index, which reflects the development of AI
and rule of law more accurately, objectively, and comprehensively in a quantifiable
and visual form, to provide support for AI and rule of law research and to promote
the development of AI and rule of law.
As a result of multidisciplinary research, the Index integrates the basic theories of
many disciplines, such as law, sociology, and statistics. The development of the Index
adheres to the concepts and principles of being people-oriented, reliable, control-
lable, standardized and orderly and striving for security, innovative development,
and shared results. It follows six major paths of AI and rule of law development,
including regulatory and policy documents, rule of law theory support, law enforce-
ment implementation effect, legal market scale, social concerns and expectations,
and rule of law environment cultivation. Developed using the Delphi method, the

ix
x Speech at the Release of the Index and the Evaluation Report

Index underwent several rounds of design and revision. Through in-depth research,
extensive consultation, and several rounds of expert verification, the expert team
sought the opinions of 43 authoritative experts in law, science and technology, and
finally determined six primary indicators, namely, policy documents, rule of law
theory, law enforcement, legal market, social concerns, and legal environment, and
18 secondary indicators, such as regulations and policies, judgments and rulings,
and determined the indicator weighting. Since the evaluation indicators are not all
scored by a percentage system, they cannot be simply added up in the comprehensive
scoring, and the data of different units need to be synthesized into a total index score.
Therefore, it is necessary to use statistical dimensionless processing to convert the
actual value of the index into the assessed value of the index, and on that basis, a
weighted sum method is used for the comprehensive scoring of the index. Combined
with the level of AI and rule of law development, the level of the specific evaluation
object is determined.
The AI development index is conducive to improving the pertinence, systematic-
ness, and authority of AI legal construction planning layouts, integrating AI devel-
opment and application into the rule of law track, and using the rule of law to guide,
standardize, and guarantee the safe, reliable, controllable, and healthy development
of AI.

On the Evaluation Report on the Development of AI and Rule

of Law in China (2019)

According to the evaluation index system and assessment methods set by the Index,
the Blue Book Editorial Board organized a team of experts to assess the development
of AI and rule of law for the first time. As the first assessment of AI and rule of law, this
assessment is exploratory. Therefore, this assessment attempts to start with simple
aspects and gradually increase the difficulty. We set the scope of this assessment in
China. This data sampling was evaluated using both questionnaires and objective
assessments to collect relevant data to evaluate the development of AI and rule of
law in China in 2019.
The assessment was launched in March 2020 with the following main methods:
first, questionnaires. The evaluation team conducted questionnaires for four groups:
rule of law researchers, users of AI legal products, technicians, and other categories of
personnel, and a total of 1,444 valid questionnaires were collected. Second, objec-
tive measurement. According to the AI and rule of law assessment index system,
objective data in multiple places (mainly through the Internet) were collected and
aggregated for measurement, and data testing was performed to ensure the objectivity
and accuracy of the statistics.
According to the evaluation data, the overall score of AI and the rule of law
in China in 2019 was 83.79, which reflects the rapid development of AI in China.
Speech at the Release of the Index and the Evaluation Report xi

Technological progress has brought about a revolutionary change in ideas, technolo-

gies, and business models. The market scale has been expanding continuously. The
promulgation of relevant laws and policy documents has been increasing year by
year. Legal research has paid close attention to practical development needs. The
AI legal environment has been gradually optimized. This has greatly changed the
traditional legal thinking mode, working method, and operation mode. However, at
the same time, there are also hidden worries in the development of AI and rule of
law: first, imbalance. It is evident that the closed-loop circle formed by the evaluation
index is characterized by nonperiodicity. Second, heterogeneity, that is, the scores
of subjective and objective indexes are quite different. Third, unevenness and the
development of AI practice have been far ahead of theoretical research, thus forming
a differential pattern of AI and rule of law. Fourth, the research on AI and rule of
law theory is directly proportional to the development of the AI industry, that is,
municipalities directly under the central government and eastern and central regions
are more prosperous in both aspects (valid questionnaires are mainly collected in
Shanghai, Beijing, Chongqing, Tianjin, Guangdong, and Henan).
The results of this assessment are a more objective and realistic reflection of the
positive results achieved in the development of AI and rule of law in China. However,
as this is the first evaluation, there is no precedent and no experience to follow, and
there are many deficiencies in index setting, evaluation methods, data possession,
and public awareness and participation, which need to be improved and enhanced
in the future. It is very much hoped that the participation and support of experts and
scholars from the legal and scientific communities and the general public will be
obtained to jointly contribute to the promotion of the development of AI and rule of
law systems and relevant research.

Liu Xiaohong
Deputy Administrator of the Editorial Board of the Blue Book
Expert Evaluation Team of the Blue Book
July 2020
An Overview of the Development of Artificial
Intelligence and Rule of Law

The COVID-19 spread throughout the world in 2020 has taken a heavy toll on
human society and the global economy. In the battle against COVID-19, artificial
intelligence (AI) has played an extraordinary role in safeguarding the common home
of humankind. A series of AI technologies have been widely applied on the frontline,
such as health QR codes, whole genome sequencing, CT imaging, and smart medical
care. Their presence to provide support in each area, including epidemic monitoring
and analysis, virus traceability, prevention, control and treatment, and resource allo-
cation, has significantly improved the capability and efficiency of epidemic preven-
tion and control. With its extraordinary performance in the fight against the virus,
AI seems to herald a new period of explosive growth. People have had a deeper
understanding of the power and role of AI.
As a “double-edged sword”, AI has brought risks and challenges while empow-
ering society. Increasing attention and concern from the public has been aroused for
increasingly prominent problems involving security, law, ethics, and other aspects in
the rapid development of new-generation AI.
Early on September 17, 2018, President Xi Jinping, in his congratulatory letter
to the 2018 World Artificial Intelligence Conference, pointed out that “Seizing this
opportunity to address the new issues raised by AI in the areas of law, safety, employ-
ment, ethics, and governmental governance requires deeper cooperation and joint
discussions among countries”.
How can we respond to and deal with these new issues of risks and challenges
posed by AI? In response, the past 2 years have witnessed the emergence and deep-
ening of research on ethics, technical specifications, industry safety, and personal
privacy protection, as well as legislation, law enforcement, justice, and other legal
problems relevant to AI. It is against this backdrop that the new concept and philos-
ophy of AI and rule of law came into being. Rapid AI development has promoted
the construction of AI in rule of law, which in return provides a guarantee for safe,
reliable, controllable, and healthy AI development.

xiii
xiv An Overview of the Development of Artificial Intelligence and Rule of Law

Fig. 1 Industry scale and growth rate of the new brand generation of AI. Sources: Chinese Institute
of Electronics, ASKCI Consulting Co., Ltd.

I. AI Development (2019–2020)

The industry scale of the new-generation AI broke 70 billion U.S. dollars in 2019
and is expected to reach close to 100 billion U.S. dollars by 2020, with an average
growth rate of 31.6 percent each year from 2018 to 2022.1

1. AI Development in China

After the upsurge in 2019 to 2020, the AI industry in China embraced a new round
of strategic opportunities, as there is an increasingly maturer condition of data, algo-
rithms, and computing power for the advancement of AI technologies and industry
and a good momentum in intelligent chips and drones, smart connected cars, intelli-
gent robots and other industry segments, as well as in medical health, finance, supply
chains, transportation, manufacturing, rail traffic, and other key areas.

1An ASKCI Consulting Co., Ltd: Research Report on the Market Outlook and Investment
Opportunities in China’s Artificial Intelligence Industry (2020–2025).
An Overview of the Development of Artificial Intelligence and Rule of Law xv

1.1 Technology Innovation: Increasingly Mature Conditions of Data,

Algorithms, and Computing Power in the AI Ecosystem

(1) Data. The rapid development of emerging tech industries such as 5G, the Internet
of Things (IoT), and automotive electronics and accelerated 5G network deploy-
ment have generated exponential and centralized increases in the enormous
volume of data from 2019 to 2020.
(2) Algorithms. Vision and tactile sensor-based transfer learning, various autoen-
coders (VAEs), and generative adversarial networks (GANs) have emerged as
new types of algorithms in unsupervised learning and have been rendered more
efficiently in model libraries of mainstream machine learning algorithms from
2019 to 2020.
(3) Computing power. The main hardware bearer of China’s AI computing power is
iterating from GPU chips into ASICs, FPGAs, and other computing unit types
in 2019 to 2020. The latter has started to be an underlying hardware capability
in support of China’s AI technology development.

1.2 Policy Orientation: Announcement of AI Development Plans

and Establishment of AI Innovation Application Pilot Area

In 2019, 19 provinces in China already announced AI development plans, 16 of

which set specific growth targets for the AI industry scale. In the second half of
2019, Shenzhen City in Guangdong Province and “Ji’nan—Qingdao” after Shanghai
Municipal set up the first Artificial Intelligence Innovation Application Pilot Area in
May were approved to do so in China, providing new opportunities for AI industry
applications. By 2020, China as a whole will focus on the “Pilot Zone” to further
promote the deep integration of the AI industry and the real economy.

1.3 Impact of COVID-19: Essential Role of AI Technology in Epidemic

Control and Prevention

The essential role of AI technology in epidemic control and prevention should be

given full play, pointed out in the Initiative to Fully Leverage AI for Empowering
Human Beings to Defeat COVID-19 in Collaboration by the Ministry of Industry
and Information Technology of the People’s Republic of China on February 4,
2020. During epidemic control and prevention, the wide application of AI tech-
nologies substantially improved work efficiency. It is believed that the AI industry
will embrace rapid growth after COVID-19.
xvi An Overview of the Development of Artificial Intelligence and Rule of Law

Table 1 Content of Initiative to Fully Leverage AI for Empowering Human Beings to Defeat
COVID-19 in Collaboration
Number Content
1 No time should be lost in applying the AI technology to reinforce weak links in
epidemic management by accelerating the advancement of industrial production and
application services. Guided by the demand, AI enterprises and application units, as
well as upstream and downstream firms are encouraged jointly to achieve
breakthroughs with positive and due contributions to epidemic detection, warning,
prevention and treatment, and other aspects
2 Efforts should be made to fully tap the potentiality of application of AI in diagnosis
and treatment and epidemic prevention and control. There should be breakthroughs
and mass production in products aiding diagnosing, fast testing, intelligent equipment,
accurate temperature measuring, target identification, and other products alike, to
facilitate AI-assisted diagnosis and treatment of diseases, reduce the risk of infection
among medical staff, and improve efficiency
3 A strong guarantee should be provided to keep daily life and work in order during the
epidemic. The initiative of “working from home, no suspension of classes and
production” has been launched with an open access to telework, video conference
services, and AI teaching resources to enable work, education, and production in a
remote, online, and intelligent way
4 The AI algorithms and computing power should be optimized to help the R&D and
breakthroughs in virus genome sequencing, vaccines/medicine, protein screening, and
other drugs

2. AI Development in Some Countries and Regions

of the World2

2.1 US: Deployment for AI Future Development in an Active Way

In February 2019, President Trump signed Executive Order 13859 announcing the
American AI Initiative. Later, in June, the Select Committee on Artificial Intelligence
of the White House Office of Science and Technology Policy issued the National
Artificial Intelligence Research and Development Strategic Plan: 2019 Update. The
influence of AI on national security takes a prominent position in Summary of the 2018
Department of Defense Artificial Intelligence Strategy: Harnessing AI to Advance
Our Security and Prosperity by Department of Defense on February 12, 2019. In the
Budget of Fiscal Year 2019, the United States has made AI a priority in its budget
and planning.

2 This part is a combination of 2019 Global Artificial Intelligence Strategy and Policy Observation
by the China Academy of Information and Communications Technology (CAICT) and Artificial
Intelligence Industry Alliance, Innovation Research by National Academy of Innovation Strategy,
and other relevant materials.
An Overview of the Development of Artificial Intelligence and Rule of Law xvii

2.2 EU: Joint Advancement of AI Development

In January 2019, the EU established the AI4EU consortium to build the first European
Artificial Intelligence On-Demand Platform and Ecosystem to share AI resources
and facilitate collaboration between all European actors in AI, with joint efforts
to advance AI development. In February 2020, the EU released digitalization and
data strategies and white papers to improve its comprehensive competitiveness in
the digital economic field. In the Ethics Guidelines for Trustworthy AI prepared
by the High-Level Expert Group on AI in April 2019, AI development based on a
people-oriented approach should be ensured on the path to benefit individuals and
society.

2.3 UK: Promotion of Continued and Healthy Development in the AI

Industry

In February 2020, the UK Committee on Standards in Public Life published its

Artificial Intelligence and Public Standards: Report to establish the regulatory and
governance framework for AI in the public sector. In June 2019, the British govern-
ment announced investing 18.5 million pounds in support of enhancing AI and data
science training.

2.4 Germany: To Take a World’s Leading Position in the AI Field

In February 2019, the German Federal Ministry for Economic Affairs and Energy
released National Industrial Strategy 2030 (Draft), clearly stipulating “to promote
AI applications and make Germany one of the world’s leading sites in AI research,
developing and applications”. AI was listed as one of the key areas to make an invest-
ment in the government draft for the 2019 Federal Budget. The German Data Ethics
Commission made policy recommendations on regulating data and AI algorithms by
law.

2.5 Japan: To Build a Powerful AI-Empowered Country

Japan takes a leading position in R&D and the application of intelligent robots
worldwide. The Japanese government rolled out its AI Strategy 2019 to strengthen
top-level design and strategic guidance in June 2019. Meanwhile, more investments
were made in AI R&D and talent training in the 2019 budget summary.
xviii An Overview of the Development of Artificial Intelligence and Rule of Law

2.6 ROK: To Improve Competitiveness in the AI Field

The Korean government announced the National AI Strategy in December 2019,

planning to improve its competitiveness in the AI field and upgrade to a world-leading
level by 2030.

2.7 Russia: Clear Statement of AI Development Strategy

Russia set up specific tasks for AI in the coming decade in the National Strategy
for the Development of AI for the Period until 2030, which was adopted in October
2019.

2.8 Singapore: National AI Strategy as the Next Key Frontier

In November 2019, the Singaporean government rolled out the National Artifi-
cial Intelligence Strategy, spelling out goals, methods, and key plans in future AI
development, as well as anchoring a vibrant and sustainable AI ecosystem.

II. AI Development in Rule of Law (2019–2020)

1. Overview of Development in AI Policies, Rules,

and Regulations

1.1 Policy Planning from Macro-deployment to Specific Measures

From AI policies, major countries and regions in the world have gradually transi-
tioned from deploying macro-planning in 2016 to 2017 to implementing specific
measures in 2019 to 2020.
China. All ministries and committees of the State Council and provincial govern-
ments have introduced and implemented policies step by step in industry, scientific
research, education, environment, intelligent cars, and other specific areas.
Other countries and regions in the world. As the national macro-planning, the US
issued the Executive Order on Maintaining American Leadership in Artificial Intel-
ligence, then with specific measures implemented in scientific research, intelligent
driving, and other areas. Russia proposed actions to advance application promotion
and raise public awareness of the National Strategy for the Development of AI for the
Period until 2030, which was adopted in October 2019. In February 2020, the EU
published the White Paper on Artificial Intelligence—A European Approach to Excel-
lence and Trust to facilitate the implementation of specific policies and measures to
promote the development of artificial intelligence.
An Overview of the Development of Artificial Intelligence and Rule of Law xix

1.2 Prioritizing Institutional Norms and Principles, Prudence

in Regulations

Considering that AI development is still in its initial period, in general, the major
countries and regions in the world are not eager to introduce regulatory measures but
to formulate a framework of governance principles.

(1) China. In the governance principles for developing responsible AI adopted in

June 2019, recommended guidelines were made from eight aspects, namely,
harmony and being human friendly, fairness and justice, inclusion and sharing,
respect for privacy, safety and controllability, shared responsibility, open and
collaboration, and agile governance.
(2) Other countries and regions in the world. The US issued the draft of Guiding
Principles for AI application to solicit public opinions in January 2020. The
EU announced the Ethics Guidelines for Trustworthy AI in April 2019. Later,
in October, Russia, in its national AI strategy, added seven basic governance
principles, including liberty of human rights, safety and reliability, transparency
and fairness, technology sovereignty, innovation and collaboration, resource
conservation and market competition. To date, except for intelligent car testing
and other relatively mature AI application scenarios, few laws or regulations
have been enacted to regulate AI.

1.3 Emphasis on Open Data and Other Government Resources

Policy planning and rules and regulations in all major countries and regions around
the world pay high attention to open data and other government resources.
China. In September 2019, Shanghai issued the Action Plan for Building Shanghai
into an AI Hub with a World-Class Innovation Ecosystem from 2019 to 2021;
in October, Beijing announced the Work Plan to Boost AI Industry Development
Through Opening Public Data. These factors lead local governments to open their
data resources.
Other countries and regions in the world. Measures to create and upgrade various
open government resources are involved in policies, such as America’s Federal Data
Strategy 2020 Action Plan in December 2019, the European Data Strategy in February
2020, and Russia’s National Strategy for the Development of AI for the period until
2030 adopted in October 2019.

1.4 Focus on Education, Fundamental Research, and Infrastructure

in the AI Field

Worldwide, major countries and regions attach great importance to the development
of AI education, fundamental research, and infrastructure.
xx An Overview of the Development of Artificial Intelligence and Rule of Law

China. “To leverage…the edge of AI…to build a learning society”, pointed out
in the decision on some major issues concerning how to uphold and improve the
system of socialism with Chinese characteristics and advance the modernization
of China’s system and capacity for governance adopted at the Fourth Plenary of
the 19th CPC Central Committee on October 31, 2019. At the 12th meeting of
the central committee for deepening overall reform on February 14, 2020, it was
proposed to “build a modern infrastructure system that is efficient, cost-effective,
green, intelligent, safe and reliable”. The meeting of the Standing Committee of the
Political Bureau of the CPC Central Committee said “a number of important projects
must be launched, and the construction of traditional and new infrastructures such as
5G and artificial intelligence must be sped up” on April 29, 2020. The Government
Work Report pointed out “provide stable support for basic research and application-
oriented basic research…and step up the construction of new types of infrastructure”
on May 22, 2020.
Some countries and regions in the world. The US launched the national artifi-
cial intelligence initiative by releasing the “National Artificial Intelligence Research
Institutes” program solicitation. The EU, Russia, and other countries and regions
all attach great importance to AI education and training, as well as its R&D and
innovation.

2. Promoting Further Research on AI and rule of law

2.1 More Emphasis on the Systematic Construction of the Basic Theory

of AI Law

From 2019 to 2020, scholars paid more attention to the study of AI jurisprudence
by scientifically defining the corresponding rights, obligations, responsibilities, and
other basic categories, laying solid foundations, and providing theoretical support
for the sound development of the legal system regulating AI. They also emphasize
establishing a regulatory system compatible with the development of AI with a
focus on finance, transportation, medical care, and urban construction, building a
reasonable and practicable AI law system following statutory procedures to ensure
that the development of AI obeys rules and laws. In addition, more research has been
conducted on the core of AI law, primarily on theoretical and practical discussions on
“AI” and “big data”, in which scholars analyze legal issues in AI applications from
an overall perspective, providing more systematic and in-depth theoretical research
on the connotation and the major body of accountability in AI law.

2.2 More Emphasis on the Rule of Law in AI law Enforcement

From 2019 to 2020, scholars emphasize accelerating the intellectualization of admin-

istrative decision-making and improving law enforcement efficiency with transparent
An Overview of the Development of Artificial Intelligence and Rule of Law xxi

data and fair algorithms by integrating AI into administrative law enforcement. To

realize more standardized administrative law enforcement, more efforts have been
made to formulate administrative regulations and designate the remit, power list,
and law enforcement procedures of law-enforcing departments to ensure that they
have laws to abide by in law enforcement activities of AI. Scholars also pay more
attention to the integration of AI in the judiciary system, accelerate the construction
of “Digital Court/Procuratorate” and “Smart Court/Procuratorate”, and promote the
modernization of the judicial system and judicial capacity.

2.3 More Emphasis on Balancing Government Power Boundaries

and Individual Privacy Protection

From 2019 to 2020, scholars have fully realized that the efficient use of AI would
generate tremendous value for social governance. However, the realization of this
value still requires system-level regulation, of which privacy protection and govern-
mental authority regulation are only two. Scholars try to avoid technology-obsessed
data centrism, which values technology most and situations where capital and power
dominated big data. Instead, they respect human dignity, individual freedom, and the
rule of law to maximize AI’s value for humankind’s benefit by managing technology’s
application situations. Existing research hopes that through effective governance of
the specific contexts in which technology is used, we can avoid the data centrism
of technological supremacy, prevent the use of big data from being completely
dominated by capital and power, respect human dignity, respect individual freedom,
respect the rule of law, and ensure that it achieves maximum value for humanity.

2.4 More Focus on the International Cooperation in AI and Rule of Law

From 2019 to 2020, scholars have explored international cooperation in AI and the
rule of law on administrative law enforcement, criminal liabilities, evidence prob-
lems, privacy protection, intellectual property protection, automatic driving, legal
entity status, judicial influences, and legal impacts of blockchain technology.

3. Further Integration of AI and Rule of Law

3.1 The Shift from Individual R&D to Joint R&D

From 2019 to 2020, more high-tech companies joined in the research and develop-
ment of AI and the rule of law by promoting the in-depth application of technologies
such as big data, cloud computing, blockchain, and 5G in the rule of law. In China,
“Trial-centered Litigation Reform Software” developed jointly by the Shanghai High
People’s Court and IFLYTEK Co., Ltd. continues to be updated and iterated. In the
xxii An Overview of the Development of Artificial Intelligence and Rule of Law

U.S., high-tech companies such as Google, Amazon, and Microsoft work with the
government, universities, and research institutions for AI R&D and applications.

3.2 The Development of Comprehensive and Specialized AI Applications

From 2019 to 2020, in China, the AI-assisted rule of law was forging ahead. Both
comprehensive and specialized AI applications are flourishing. In Western countries,
AI was mostly used to process legal documents in the judicial field, while specialized
applications are also on the rise.

3.3 Applications: From Stages to the Whole Process

From 2019 to 2020, in China, AI gradually covered all aspects of the rule of law,
building a convenient, efficient, and deeply integrated AI and rule of law system. In
the United States, AI has been used in multiple judicial situations, including criminal
investigation and sentencing.

3.4 Data: From Isolation to Integration and Sharing

From 2019 to 2020, the fact that data have not yet highly converged and shared is still
one of the challenges that hinder the development of AI applications in rule of law, but
platforms of AI application in rule of law in various fields have been exploring data
sharing models across levels, geographies, departments, and businesses. In China,
19 provinces have established provincial big data resource bureaus (centers). In the
United States, the Office of Management and Budget (OMB) released the Federal
Data Strategy and 2020 Action Plan on December 23, 2019, proposing a “sharing
initiative” for open databases and cross-institutional resource sharing.

4. New Features of AI-related Cases

4.1 Explosive Growth of AI-related Cases

McKinsey & Company interviewed 2,360 employees from various industries world-
wide in 2019. The survey showed that 58% believed their companies were using
AI technologies in at least one business and management activity, which was 11%
higher than that in 2018. As technology matures and the underlying technical frame-
work becomes open source, the industry continues to grow at an accelerating rate,
driving a rapid increase in the number of AI-related cases worldwide. According to
incomplete statistics, from June 2018 to June 2019, there were 116 AI-related cases
An Overview of the Development of Artificial Intelligence and Rule of Law xxiii

in China, and the rapid development of the AI industry led to a significant increase
in the number of AI-related cases, with 371 cases from July 2019 to June 2020.

4.2 Uneven Geographic Distribution of AI-related Cases

According to the 2019 Stanford University Artificial Intelligence Index Report,

geographically, AI-related cases are concentrated in cities with more developed
industries, such as finance, electronics technology, and manufacturing. According to
the statistical analysis, AI-related cases in China are mainly found in the eastern and
central parts of the country, with Beijing, Guangdong, Hubei, Jiangsu, and Shanghai
accounting for 70.02% of the total number of AI cases.

4.3 Increasing Social Awareness of AI-related Cases

Currently, the global community is paying more attention to AI. The new types of
cases caused by the application of AI have brought unprecedented impact and chal-
lenge to the established social ethical standards, legal rules, social order, and public
management system. For example, the so-called “First Case of Facial Recognition”
in China in 2019 is a classic example of the conflict between personal privacy and
technological development in the era of big data.

4.4 Global and Covert AI-related Cases

Since its inception, crimes involving AI have been breaking through the boundaries
of the national field and have been spreading and radiating in the form of networks.
Currently, crimes have become global in nature, which means that the national gover-
nance of AI should be expanded to a global level. All countries should work together
on it.

5. “AI & Law” as a Discipline Bears Fruit

5.1 Serving the National Strategy and Regional AI Hub

In China. The State Council defined a “three-step” strategy in the Development

Plan for a New Generation of AI (July 20, GF [2017] No. 35). According to the
plan, Beijing, Shanghai, and other leading cities have been working on a new AI
development system featuring application-driven, technology-led, collaborating with
industries, valuing ecological cultivation and talent clustering, taking the lead in
creating AI hubs in China, and cultivating interdisciplinary talents proficient in AI
and the rule of law.
xxiv An Overview of the Development of Artificial Intelligence and Rule of Law

In other countries and regions. The United States, the United Kingdom, Germany,
France, Japan, South Korea, Canada, the European Union, and other countries and
regions all value fundamental AI research and personnel training and are devoted to
increasing investments in those fields.

5.2 Innovating “AI & Law” Discipline

In China. To achieve the deep integration of “AI and law”, the major of AI and
law in Chinese universities is establishing a professional comprehensive laboratory
to support its teaching; establishing student internship bases covering the court,
procuratorate, public security system, law firms and well-known AI enterprises; and
developing specialized teaching materials for this major. Efforts have been made to
actively explore the in-depth integration of disciplines and innovative talent training
models and cultivate new academic growth points. For example, Shanghai Univer-
sity of Political Science and Law established the “School of AI and Law” in 2019
and enrolled 40 undergraduate students that year, which marked the beginning of
undergraduate education in AI and Law.
In other countries and regions. AI Law is highly valued since it is an emerging and
distinctive discipline. Many overseas law schools, led by some prestigious univer-
sities in the United States, have opened AI and law courses and established AI
law research institutes. For example, Stanford University of the US has opened an
interdisciplinary course of law and computer science.

5.3 Cultivating Interdisciplinary Talents Proficient in AI and Law

As a “double-edged sword, while bringing higher business value, more jobs, and
better lives, AI also has drawbacks, such as posing threats to human life, creating
challenges to information security and even national security, and having impacts on
law and social ethics. In facing potential AI risks, we should make plans in advance.
We should speed up the cultivation of interdisciplinary talent proficient in AI and
the rule of law, actively learn and use professional knowledge and necessary means
to predict the future of our AI technology. To ensure AI’s healthy development, we
should also study issues such as safety, responsibility, equality, and quality of life to
establish a feasible legal system regulating AI in the twenty-first century.

III. The Prospect of AI and Rule of Law

On July 20, 2017, The State Council of the People’s Republic of China issued the
Notice of the State Council on the Issuance of a Plan for the Development of a
New Generation of AI (GF [2017] No. 35), setting a “three-step” strategic goal for
AI’s future development. The first step goal has largely been achieved: “by 2020,
An Overview of the Development of Artificial Intelligence and Rule of Law xxv

overall AI technology and its applications will keep pace with the world’s leading
countries; the AI industry will become a new important economic growth point, and
the application of AI will become a new way to improve people’s livelihood, strongly
supporting the goal of entering the ranks of innovative countries and realizing the
goal of building a moderately prosperous society in an all-round way”. From 2021,
we will work on the second- and third-step goals.
Undoubtedly, whoever has an advantage over AI will be able to gain the initiative
in the competition to win the future. To dominate the new round of international
science and technology competition, all countries see AI as an essential part when
enhancing national competitiveness and safeguarding national security. Plans and
policies on core technologies, top talents, standards, and relevant regulations have
been put in place. However, AI also has its drawbacks. It may bring risks while
empowering society. Thus, we should make plans in advance, carefully evaluating
risks that AI might bring in legal, safety, employment, moral, ethical, and govern-
mental dimensions. It also calls for global efforts to avoid risks along the way. More
people believe that we should establish a rule of law system regulating AI, using
the rule of law philosophy and methods to lead, promote, regulate, and protect the
healthy development of AI.
Technology development never ends, and so does the progress of the rule of
law. We, legal professionals, shoulder great responsibilities and undertake honorable
missions along the way. We will keep our eyes on the frontiers of AI technology and
make the most of legal research to offer “Shanghai Solution” and “China Solution” in
establishing an AI and rule of law system. We will be devoted to creating AI and rule
of law ecology; developing AI and rule of law hubs promoting the global governance
of AI; and ensuring its safe, reliable, controllable, and healthy development.
AI and the Rule of Law Development Index

Expert Group of the Blue Book on the World AI and Rule of Law
July 10, 2020

1. Introduction

AI is the cutting-edge achievement of human technology and the crystallization of

wisdom. As a modern strategic and disruptive technology, AI has been the new
core competency for future development. Moreover, recognized as a “double-edged
sword” characterized by a high integration of technology and society, AI brings
benefits as well as risks and challenges to human beings.
“A new generation AI is booming all over the world, injecting new energy into
economic and social development and profoundly changing people’s production and
lifestyle. It is necessary for all countries to deepen cooperation and explore together
to seize development opportunities in a timely manner and appropriately address the
new issues concerning law, security, employment, ethics, government governance
and other aspects in AI development”, pointed out by President Xi Jinping in his
congratulatory letter to the 2018 World Artificial Intelligence Conference held in
Shanghai on September 17.
“While robustly developing AI, we must highlight the potential safety risks,
enhance early prevention and guidance, reduce risks to a maximum degree and ensure
the safe, reliable and manageable development of AI”, stated in the Next Generation
Artificial Intelligence Development Plan issued by the State Council of China.
How can the possible risks and challenges posed by AI be handled? Perhaps the
best answer to this question is the rule of law, which serves as an efficient way to
ensure the safe, reliable, and manageable development of AI.
AI and rule of law came into being at this historic moment. As a new concept
and a new field, AI and rule of law involve legislation, law enforcement, justice and
law-abiding, and many other aspects in the AI field. Accordingly, the research on AI
and rule of law has emerged, the scope of which covers basic theories, policies and
regulations, systems and mechanisms of AI and rule of law, as well as legislation,

xxvii
xxviii AI and the Rule of Law Development Index

law enforcement, justice, law-abiding, and other aspects of AI. This is the call of our
times and an internal requirement of AI development to promote research on AI and
rule of law.
To better promote research on AI and rule of law, the expert group on the Blue Book
on AI and Rule of Law in the World compiled the AI and the Rule of Law Development
Index, containing the index system, indicator weighting, and evaluation method for
assessing the development of AI and rule of law. It aims to form a special report by
evaluating AI development in China, represented fully in a measurable and visible
way, to provide references for research and policy-making, as well as to facilitate the
establishment of a research system for standardized development and risk prevention
and control of AI at the national level and the improvement of the systemic layout and
coordinated guidance on its research. Meanwhile, this research on the Development
Index and its binding and guiding index can further enhance the influence of the
Blue Book, making positive contributions to promoting international exchanges and
cooperation in AI and rule of law, global AI governance and creating the AI and rule
of law system and ecology.
After “designing and revising” the index system, the Expert Group completed
the final version of the AI and the Rule of Law Development Index through in-depth
investigation and research, extensive public opinion soliciting procedure, and several
rounds of expert argumentation.

2. Index System

2.1 Evaluation Index System

The evaluation index system integrates law, statistics, sociology, data science,
computer science, and other relevant disciplinary theories, laying the foundation
for achieving the goal of this evaluation and leading future development. Whether
the index system is scientifically sound can affect the authority and credibility of the
AI and rule of law assessment. Under the principles of human-centredness, safety,
and innovation, this evaluation emphasizes shared achievements and aims to promote
the reliable, controllable, and orderly development of AI. Establishing the evaluation
index system and conducting objective measurements are prerequisites for various
theoretical studies. Furthermore, merits and demerits in the development of AI and
rule of law can be found in this process.
The development of AI and rule of law can be carried out mainly from six
dimensions, namely, policy documents, rule of law theory, law enforcement, the
legal market, social concern, and rule of law environment. The internal logic of the
aforementioned six dimensions enables AI and rule of law to have unique internal
mechanisms, behavior patterns, and logic.
AI and the Rule of Law Development Index xxix

Fig. 1 Diagram of the Index

System for Evaluating AI
and Rule of Law
Development

2.1.1 Policy Documents

With the advancement of big data, 5G, telecommunications, IoT, and other basic tech-
nologies, AI has been profoundly changing people’s lives and productivity. Never-
theless, apart from bringing benefits, AI also comes with risks. In response, efforts
should be made for AI risk prevention, and a legal framework for AI regulation
should be established to enable its healthy development and maintain the rule of law
and order in the AI era. Policy documents are thus an important engine to drive AI
industry development and mark the AI development direction and boundary. World-
wide major countries, while issuing their national AI development strategies and AI
industry promotion policies one after another, have begun to explore the direction
and method for AI rules and regulations. Much emphasis is placed on establishing
legislative norms, the legal system, the policy system, and ethical norms compatible
with innovative AI development, developing the legal system for AI safety assess-
ment and risk prevention and control, and enhancing the ability to manage technical
and application rules.

2.1.2 Rule of Law Theory

As the saying goes, “theory pales in comparison with the evergreen tree of practice”.
In the face of numerous uncertainties in AI development, support by the rule of
law theory is even more needed than before. Efforts should be made to identify the
xxx AI and the Rule of Law Development Index

direction and key areas in creating the national research system for standardized
development and risk prevention and control of AI, as well as in planning layout
and coordinating guidance on research to make the work more targeted, systemic,
authoritative, and professional. It is stressed to strengthen AI theoretical research for
advancement in the interactive mechanism to support technical innovation, binding
guidance and standardization in the AI field. The establishment of legislative norms,
the legal system, the policy system, and ethical norms should be compatible with
innovative AI development to develop legal mechanisms for AI safety assessment
and risk prevention. Over and above, rules and regulations for the technical design
should be formulated, which can improve the fairness, transparency, and security of
AI algorithms without algorithm discrimination and eliminate technical applications
against ethical values and public order and good morals.

2.1.3 Law Enforcement

With deep learning emerging, AI has gradually become a strong driver in industry
applications. AI, cloud computing, and big data continue to empower devices for
law enforcement, facilitating the development of law enforcement informatization
and leading the new trend in law enforcement applications. As industry applica-
tion paradigms in AI-empowered law enforcement appeared one after another, law
enforcement informatization for the new era has turned from a beautiful blueprint to
a daily work routine, while law enforcement plus AI has been a featured, profession-
alized, and scaled scenario. The deepened application of AI in law enforcement has
not only improved law enforcement technology and standardization, but also helped
to address problems such as manpower shortage and low efficiency, thus increasing
the sense of happiness and satisfaction of the people. On this basis, to further promote
deep integration of AI and rule of law, the key is to establish safety, interest, and
accountability mechanisms for sharing data to ensure the legitimate, safe data flow
and use and to advance the deep application of AI to law enforcement, justice, and
other areas to speed up the smart rule of law development and make rule of law
practices intelligent.

2.1.4 The Legal Market

In terms of product forms, applications of AI in the law field mainly render infor-
mation retrieval, lawyer matching, corporate legal support, document review, case
prediction, intelligent counseling, and other forms alike. Actors are companies
applying AI to law, including comprehensive providers of AI technology and solu-
tion plans, such as Baidu, Alibaba Group, Tecent and iFLYTEK, and companies
with professional law information platforms or legal informatization, such as Beijing
Thunisoft, Tongdahai (tdhnet.com.cn), and Wusu (itslaw.com). Among them, Beijing
Thunisoft and Alibaba Group are listed companies, and DeepMind Technologies and
AI and the Rule of Law Development Index xxxi

other startups are involved. Services are provided to parties (individuals or compa-
nies), lawyers (lawyers and their working companies, etc.), as well as judicial agen-
cies (courts and procuratorates, etc.), and arbitration and mediation agencies. AI
product services have to vary with different actors to meet their demands. Hence, it
is imperative to further improve the depth, width, and validity of AI application in
the law field, realizing product structure optimization and the iterative upgrade of the
industrial chain. The emphasis is placed on advancing the cooperation of “policies,
learning, research, industry and application”, that is, to integrate research resources
in party and government institutions, colleges and universities, research institutes,
business and institution units, sharing wisdom and experience together to promote
the AI legal market.

2.1.5 Social Concern

With the rapid growth and wide application of AI and other new technologies, revo-
lutionary changes have been brought to people’s production manner and lifestyle.
Human society in AI’s continuous development embraces new opportunities while
confronted with new challenges. As a disruptive technology exercising a wide scope
of influence, AI, due to its risks and challenges caused by uncertainties in devel-
opment, will exert a deep and long-term influence on social stability, economic
growth, ethics, rule of law construction, and even global governance. The way to
draw on advantages in avoidance of disadvantages and share benefits has been a
public concern. What is important is to carry out research on application scenarios
of AI in rule of law, to broaden the horizon of research on preventing and punishing
crimes that implicated AI and to establish the mechanism for risk prevention and
response in complicated AI application scenarios to save AI from application to
illegal conduct. In the long run, what would happen to society if we successfully
created powerful AI that could outperform human beings in terms of cognitive func-
tion? From this perspective, following AI follows humankind’s own development.
Inevitably, it is relevant to everyone.

2.1.6 Rule of Law Environment

Fostering a favorable environment for AI and rule of law requires the initiative to
embrace new technologies. More attention should be given to potential risks and
challenges, responding to social concerns and actively advancing to build future AI
and rule of law systems to regulate the development of AI applications. Efforts should
be made to promote, regulate, and ensure the healthy and sustained development of
AI to serve the implementation of national strategies and benefit mankind. It is
recommended to expand the teaching content of AI and other majors on the basis of
law discipline and attach great importance to the integration of AI and legal education
for cultivating interdisciplinary talent with expertise both in law and AI technology.
xxxii AI and the Rule of Law Development Index

2.2 Components of the Index System

To present completeness of policy documents, contribution by theory innovation,

satisfaction with law enforcement, the steady growth rate of the market scale, the
intensity of social concern, and optimization of rule of law environment, under
the principles of science, comprehensiveness, hierarchy, and dynamic, this research
formulated the Index at three levels, namely, goals, principles, and indicators. It
contains 6 level 1 indicators and 18 level 2 indicators.
Level 1 indicators. Indicators in the goal aspect are based on the development of
AI and rule of law, measuring the development process. This evaluation indicator
can reflect from the time dimension the level and evolution of AI and rule of law
development and from the spatial dimension its overview, distribution characteristics,
inner potential, and development strength. Indicators in this aspect are made up of
six dimensions: policy documents (A1), rule of law theory (A2), law enforcement
(A3), the legal market (A4), social concern (A5), and the rule of law environment
(A6).
Level 2 indicators. Level 2 indicators are comprised of factors reflecting level 1
indicators. In other words, level 2 indicators with further refinement are the classi-
fication and synthesis of level 1 indicators. There are 18 level 2 indicators. Among
them, policy documents (A1) contain 2 level 2 indicators, namely, rules and poli-
cies (B1) and judgments and verdicts (B2); rule of law theory (A2) four, national
projects (B3), provincial and ministerial projects (B4), forum discussion and policy-
making consultation (B5) and journal publication (B6); law enforcement (A3) four,
policy implementation (B7), software running (B8), smart justice (B9) and privacy
protection (B10); the legal market (A4) two, the number of companies (B11) and
the ability to generate revenue (B12); social concern (A5) three, legislative concern
(B13), policy concern (B14), and forum concern (B15); rule of law environment
(A6) three, inclusiveness (B16), legal regulations (B17) and application experience
(B18). Of these indicators, eight are objective indicators, and the remaining ten are
subjective. These indicators can unfold the real situation of AI and rule of law devel-
opment as well as the general trend and development direction of AI change and
evolution from the dynamic side.

3. Indicator Weighting

The Index determines the relationship of indicator weight by the Delphi method. The
so-called Delphi method, also known as the expert specified procedure investigation
method, is a back-to-back communication technique to ask a panel of experts for their
forecasting answers to a preprepared questionnaire in several rounds where experts
will revise their forecasts based on other experts. After rounds of repetition, it is
believed that as the answer will gradually decrease with the group converging toward
the “correct answer”, a collective judgment result of high accuracy can be obtained.
By the Delphi method to objectively determine the weight of each indicator, one of
AI and the Rule of Law Development Index xxxiii

the key points is to select a list of suitable experts. In this respect, special attention
should be given to expert authority and the reasonable mix of members in the panel.
In the list, there should be workers taking long-term practices in the AI and rule of law
field and experts and scholars conducting relevant theoretical research in universities
or research institutes. For selection, the common requirement is that they should
understand and familiarize with the status quo of AI in rule of law construction
to a certain extent. The evaluation expert panel comprises 43 experts and leaders
from the Shanghai Municipal court system, the Shanghai Municipal procuratorate
system, Shanghai Legal Society, Shanghai Law Society, Shanghai Academy of Social
Sciences Institute of Law, Shanghai University Law School, Shanghai University of
Political Science and Law School of Artificial Intelligence and Law, and others.
Based on their opinions, the final indicator weight to assess the AI and rule of law
development was determined.

4. Evaluation Method

According to evaluation goals and requirements, the evaluation of AI and rule of law
development mainly relies on objective data of public opinions and subjective data
based on data evaluation. These two collected types of data will be used to assess
the AI and rule of law development level.

4.1 Subjective Data

The questionnaire was drawn in accordance with the evaluation indicator system
centered on the content of the system. The collection of subjective data with opinions
of different groups on AI and rule of law is completed by distributing and collecting
questionnaires in a science-based manner.

4.1.1 Questionnaire for Users

Users of AI-powered products for the rule of law are those in relevant depart-
ments of justice and law enforcement, including judges/procurators (assistant
judges/procurators), police officers, administrative law enforcement workers,
lawyers, and corporate counsel. The foregoing group has a straightforward judg-
ment and rating for the development of AI and rule of law. In terms of AI-powered
products for the rule of law, they have their own understandings and ideas about their
problems and propose different corresponding solutions, or even they can offer their
own needs of “product services”. Undoubtedly, it has positive meaning to collect and
conclude this information for improving AI and rule of law development. Regardless
Table 1 Index System and Weighting for Evaluating AI and Rule of Law Development
xxxiv

Level 1 Level 2 Objective/Subjective Testing Points Evaluation Criteria

Policy Documents A1 (19.45%) Rules and regulations B1 Objective Number of issued rules and Increase or decrease in the
(49.16%) regulations number of rules and regulations
centered on and involved AI
Judgments and verdicts B2 Objective Number of judicial cases Increase or decrease in the
(50.84%) implicating AI number of judicial cases
implicating AI
Rule of Law Theory A2 National projects B3 (29.54%) Objective Number of national projects Increase or decrease in the
(13.58%) number of law projects by
National Social Science Fund
of China (National Social
Science Fund of China, Law)
Provincial and ministerial Objective Number of provincial and Increase or decrease in the
projects B4 (22.23%) ministerial projects number of AI provincial and
ministerial projects (Ministry of
Justice, China Law Society and
Ministry of Education)
Forum discussion and Objective Number of forums (times) Increase or decrease in the
policy-making consultation B5 number of AI and rule of law
(21.74%) forums
Journal publication B6 (26.49%) Objective Number of published Increase or decrease in the
articles (pieces) number of articles on AI and
rule of law in law and other
comprehensive journals
(continued)
AI and the Rule of Law Development Index
Table 1 (continued)
Level 1 Level 2 Objective/Subjective Testing Points Evaluation Criteria
Law Enforcement (19.09%) Policy implementation B7 Subjective Effect of policy Satisfaction with the
(25.12%) implementation implementation effect of policy
and legal documents for AI
development
Software running B8 (23.07%) Subjective Effect of software running Satisfaction with the frequency
operation of software running errors in
the application of AI and rule of
law and the algorithm
discrimination
Smart justice B9 (27.91%) Subjective Level of judicial application Satisfaction with the
of AI application of AI to the justice
AI and the Rule of Law Development Index

Privacy protection B10 (23.90%) Subjective Level of privacy protection Satisfaction with exposure of
privacy in the AI-empowered
law enforcement
The Legal Market A4 (17.09%) Number of companies B11 Objective Increase or decrease in the Number of companies with “AI
(45.72%) number of companies plus rule of law” as core
business
Ability to generate revenue B12 Objective Increase or decrease in the Company’s ability to generate
(54.28%) ability to generate revenue revenue with “AI plus rule of
law” as core business
Social Concern A5 (17.35%) Legislation concern B13 Subjective Attention to legislation to Attention to introducing
(35.14%) take a step ahead regulations for illegal activities
with the use of AI and
committed by the AI system
Policy concern B14 (33.19%) Subjective Attention to policy Attention to policy and legal
documents documents of AI development
(continued)
xxxv
Table 1 (continued)
xxxvi

Level 1 Level 2 Objective/Subjective Testing Points Evaluation Criteria

Forum concern B15 (31.67%) Subjective Attention to rule of law Attention to relevant forums
forum with “Rule of Law Forum of
World Artificial Intelligence
Conference” held in Shanghai
as representative
Inclusiveness B16 (33.56%) Subjective Inclusiveness of AI Inclusiveness for AI
development
Legal regulations B17 (32.09%) Subjective Inclusiveness of AI Legal inclusiveness for overall
AI development
Rule of Law Environment A6 Application experience B18 Subjective AI application experience Experience of AI applications
(13.44%) (34.35%)
AI and the Rule of Law Development Index
AI and the Rule of Law Development Index xxxvii

of lawyers, corporate counsel, or participants in developing such products, adminis-

trative counterparts, litigants, or ordinary observers, each of them experiences and
goes through each step in the AI and rule of law development and finally makes their
own judgment. This is why product users are the respondents to the questionnaire.

4.1.2 Questionnaire for Technical Workers

AI technical workers and other relevant technical workers, AI majors and other
students in relevant majors were the second type of respondents. This group directly
participates in the design and R&D of AI products for the rule of law and later serves
as a service provider. As they have a relatively deeper understanding and experience
of the current situation, inner rules and potential problems of AI and rule of law
development, they can make relatively more professional comments on AI in rule of
law work.

4.1.3 Questionnaire for Researchers

For researchers in AI and rule of law, their work determines that they have to make
frequent contact with all levels of society and governments at all levels, as well as
judicial agencies. They throw themselves into disciplinary research while proposing
opinions and suggestions for policy-making consultation in order to build the think
tank. Within the scope of professional behavior norms, researchers are allowed to
have independent ideas, actions, and liability. As an important force in improving
the development of AI and rule of law, researchers maintain an inherent tie with AI
while having relatively more engaged experience with AI and rule of law work.

4.1.4 Others

In addition to the above three groups, there were other respondents, including staff of
administrative organs, judicial administrative staff, personnel of institutional units,
workers of social organizations, judicial assistants, company clerks, social workers,
retirees, mediators, and workers in villages and residents’ committees. They are
concluded as respondents because more detailed trivial knowledge of the AI and rule
of law development can be obtained through a broadened horizon.

4.2 Objective Data

With the evaluation indicator system at its center, the established objective database
for evaluating the development of AI and rule of law contains the data information
of indicators, including rules and regulations (B1), judgments and verdicts (B2),
national projects (B3), provincial and ministerial projects (B4), forum discussion and
xxxviii AI and the Rule of Law Development Index

Table 2 Levels of AI and Rule of Law Development

AI and Rule of Law Score Ranges of AI and Rule of Levels of AI and Rule of Law
Development Law Development Development
I 90 < A ≤ 100 High
II 80 < A ≤ 90 Relatively high
III 70 < A ≤ 80 Average
IV 60 < A ≤ 70 Relatively low
V 0 < A ≤ 60 Very low

policy-making consultation (B5), journal publication (B6), the number of companies

(B11) and the ability to generate revenue (B12), collected on PKULAW. COM, China
Judgments Online, TianYanCha.com, Qcc.com, National Social Science Fund of
China, Ministry of Justice of the People’s Republic of China (PRC), China Law
Society, the online database of the Ministry of Education of PRC, cnki.net, and other
channels.

4.3 Evaluation Index

The following is the formula for calculating the Index to evaluate the AI and rule of
law development:


18
A= Wi Di
i=1

Of the formula,
A—index of AI and rule of law development;
Wi—combined indicator for the i-indicator;
18
Di—the scores of the i-indicator 
i=1
Based on the research materials, the evaluation scores of indicators at each level
calculated by the aforementioned formula can be concluded. A further step to grade
the comprehensive index into different levels can determine the level of the AI and
rule of law development. There are five levels, namely, high, relatively high, average,
relatively low, and very low. In this way, the current level of development can be
determined.
Assessment Report on Artificial Intelligence
and the Rule of Law in China (2019)

Advisory Board of the Blue Book on AI and Rule of Law in the World
July 10, 2020
To promote the development of AI and the rule of law system, the Blue Book on AI
and Rule of Law in the World compiled the AI and the Rule of Law Development
Index (hereinafter referred to as the Index). The Index aims to reflect the development
of AI and the rule of law in China in an objective, accurate, and comprehensive way
and drive and promote AI and rule of law research. Based on the index system and
evaluation methods set by the Index, the editorial board of the Blue Book invited a
team of experts to evaluate the development of AI and the rule of law in China 2019.
As the first assessment of AI and rule of law, this assessment is exploratory. Therefore,
this assessment attempts to start with simple aspects and gradually increase the
difficulty. We set the scope of this assessment in China. This data sampling was
evaluated using both questionnaires and objective assessments to collect relevant
data.

1. Overview

1.1 Subjective Data

In March 2020, we distributed questionnaires in Shanghai to four groups of people: AI

legal product users, technicians, researchers, and other relevant persons. In addition,
the Shanghai Law Society invited local law societies in Beijing, Henan, Tianjin,
Shaanxi, Jiangsu, Shanxi, Jilin, Guangdong, Fujian, Shandong, Hebei, Hubei,
Sichuan, Chongqing, Heilongjiang, Gansu, Anhui, Guizhou, Liaoning, Yunnan, Inner
Mongolia, and other provinces to assist in the local distribution of the questionnaires.
A total of 1444 valid questionnaires were returned as of April 28.

xxxix
xl Assessment Report on Artificial Intelligence and the Rule of Law …

Table 1 Distribution of the sample

Categories Indexes Frequency Percentage (%)
Age Below 20 43 2.98
21–30 450 31.16
31–40 571 39.54
41–50 241 16.69
51–60 124 8.59
60+ 15 1.04
Occupations AI legal product Judges/Prosecutors 335 23.20
users (Including Side Judges,
Assistant Prosecutors)
Police Officers 176 12.19
Administrative 158 10.94
Law-enforcement
Officers
Attorneys, Corporate 167 11.57
Counsel
Technicians AI Technicians 8 0.55
Technicians Students major in AI and 16 1.11
related subjects
Researchers Students major in law 190 13.16
Researchers Teachers and researchers 79 5.47
Other relevant Other relevant people 315 21.81
people
Organizations Courts The Supreme People’s 2 16.14
Court
Higher People’s Courts 4
Intermediate People’s 3
Courts
Primary-level People’s 224
Courts
Procuratorates Provincial-level People’s 3 11.15
Procuratorates
Prefecture-level People’s 31
Procuratorates
Primary-level People’s 126
Procuratorates
Special People’s 1
Procuratorates
Public Security Municipal Public 57 11.36
Agencies Security Bureau
Police Stations 107
(continued)
Assessment Report on Artificial Intelligence and the Rule of Law … xli

Table 1 (continued)
Categories Indexes Frequency Percentage (%)
Universities Law-majors 202 15.65
Nonlaw majors 20
Research Institutes 4
Other relevant organizations 660 45.70

1.2 Objective Data

In April 2020, under AI and the rule of law evaluation system, we collected and
analyzed related objective data from PKULAW.com (https://www.pkulaw.com/),
China Judgments Online (https://wenshu.court.gov.cn/), Tian Yan Cha (https://www.
tianyancha.com/), Qi Cha Cha (https://www.qcc.com), The National Social Science
Fund of China (http://fz.people.com.cn/skygb/sk/index.php/index/index), Ministry
of Justice of the People’s Republic of China (http://www.moj.gov.cn/), China
Law Society (https://www.chinalaw.org.cn/), Ministry of Education of the People’s
Republic of China (http://www.moe.gov.cn/) and CNKI (https://h5.cnki.net). The
data integrity of all sources has been tested.

2. Overall Index

According to the evaluation data, the overall score of AI and the rule of law in China
in 2019 was 83.79, which reflects the rapid development of AI in China. Technolog-
ical progress has brought about a revolutionary change in ideas, technologies, and
business models. The market scale has been expanding continuously. The promulga-
tion of relevant laws and policy documents has been increasing year by year. Legal
research has paid close attention to practical development needs. The AI legal envi-
ronment has been gradually optimized. This has greatly changed the traditional legal
thinking mode, working method, and operation mode. From this perspective, the
application of AI and the rule of law has become an important driving force for the
development of smart legislation, smart law enforcement, smart justice, and smart
governance. To this end, it is necessary to build a new type of rights and obligations
relationship based on the requirements of AI development, promote the transforma-
tion and upgrading of legal rules and orders, and promote the development of AI and
rule of law in a deeper and broader field.
xlii Assessment Report on Artificial Intelligence and the Rule of Law …

Table 2 Scores and Percentage of Primary Indicators

Primary Indicators Scores Total Percentage
Policies A1 18.24 19.44 93.6
The Rule of Law Theories A2 11.16 13.58 86.3
Law Enforcement A3 13.79 19.09 72.9
Legal Market A4 16.02 17.09 95.6
Social Awareness A5 13.57 17.35 80.5
The Rule of Law Environment A6 11.03 13.44 85.3

3. Individual Indicators

3.1 Primary Indicators

For the primary indicators, the scores are as follows (in descending order): Legal
Market A4 (95.6) > Policies (93.8) > Rule of Law Theories A2 (86.3) > Rule of
Law Environment A6 (85.3) > Social Awareness A5 (80.5) > Law Enforcement A3
(72.9).
Evidently, with the rapid development of AI in China, we have seen market
size expanding, legal documents increasing, and more practice-oriented legal theory
research. Moreover, AI and the rule of law environment have been upgraded, and
society as a whole has greater expectations of the application of AI in the rule of
law, while the application of AI products in specific scenarios in law enforcement
remains relatively weak.

3.2 Secondary Indicators

For the Secondary Indicators, the scores are as follows (in descending order):
The Number of Corporations B11(95.3)≥ Journal Publications B6(95.3)> Laws
and RegulationsB1(95.2)> Judgments and Decisions B2(92.5)> Revenue Capacity
B12(92.4)> Forum Discussions and Decision Consultations B5(89)> Users Feed-
back B18(87.7)> AI Law Legislation Awareness B13(81.1)> Public Attitudes
B16(79.9)> Legal Regulations B17(78.3)> Policy Awareness B14(77.9)> Forum
Awareness B15(75.3)> Smart Juftware Operations B8(74.3)> Policy Implemen-
tationB7(73.5)> National Level Projects B73)(73)> Smart Juvel B9)(71. 9) >
Provincial and Ministerial Level Projects B4 (71.9) > Privacy Protection B10 (69.2).
More specifically, companies in AI and law are growing in numbers and sizes; law
journals pay more attention to AI and law research; policies and regulations are more
pragmatic, targeted, and practice-oriented; the judiciary is more decisive in handling
disputes; forums focus more on the cutting edge of AI development; basic functions
are more user-friendly; the whole society holds an open but cautious attitude toward
Assessment Report on Artificial Intelligence and the Rule of Law … xliii

Table 3 Scores and Percentage of Secondary Indicators

Secondary Indicators Scores Total Percentage
Laws and Regulations B1 9.10 9.56 95.2
Judgments and Decisions B2 9.14 9.88 92.5
National Level Projects B3 2.93 4.01 73.1
Provincial and Ministerial Level Projects B4 2.17 3.02 71.9
Forum Discussions and Decision Consultations B5 2.63 2.95 89.0
Journal Publications B6 3.43 3.60 95.3
Policy Implementation B7 3.52 4.80 73.5
Software Operations B8 3.27 4.40 74.3
Smart Justice B9 3.83 5.33 71.9
Privacy Protection B10 3.16 4.56 69.2
The Number of Corporations B11 7.45 7.81 95.3
Revenue Capacity B12 8.57 9.28 92.4
AI Law Legislation Awareness B13 4.94 6.10 81.1
Policy Awareness B14 4.49 5.76 77.9
Forum Awareness B15 4.14 5.49 75.2
Public Attitudes B16 3.60 4.51 79.9
Users Feedback B17 3.38 4.31 78.3
Legal Regulations B18 4.05 4.62 87.7

AI development. However, it is evident that the closed-loop circle formed by the

evaluation index is characterized by nonperiodicity. Among the primary indicators,
two are over 90, three are 80–90, and one is below 80; among the secondary indi-
cators, five are over 90, three are 80–90, nine are 70–80, and one is below 70. The
imbalance of these indicators indicates that there are short boards in the development
of AI and rule of law, which needs to be solved urgently to promote its coordinated
development.
Meanwhile, it is also important to pay great attention to the imbalance between
indicators, that is, the theoretical research lags far behind the development of AI
application, and to pay attention to the disparate patterns that exist and will persist in
the AI and rule of law development, such as insufficient stability and operability of
software application (74.3), application of AI products in specific scenarios in law
enforcement (72.9), privacy protection in the age of AI (69.2). In this regard, it is
necessary to give full play to the function of legislation to regulate, protect, and lead
to improve efficiency, maintain order, and uphold justice.
xliv Assessment Report on Artificial Intelligence and the Rule of Law …

3.3 Objective Indicators

For Objective Indicators, the scores are as follows (in descending order): The Number
of Corporations B11(95.3)≥ Journal Publications B6(95.3)> Laws and Regula-
tions B1(95.2)> Revenue Capacity B12(92.4)> Judgments and Decisions B2(92.5)>
Forum Discussions and Decision Consultation B5(89.0)> National Level Projects
B3(73.1)> Provincial and Ministerial Level Projects B4(71.9).
The average score of Objective Indicators is 88.1.
Specifically, social capital is more sensitive to the specific scenarios and future
development of AI. AI products with good development prospects and excellent
product performance have been sought after in the capital market. Legal theorists
also pay close attention to the multiple possibilities of the future development of AI
and design various coping strategies accordingly. Moreover, the closed-loop circles
formed by objective indicators are less smooth than those of subjective indicators,
suggesting that different aspects of the development of AI and the rule of law need
to be coordinated to make its development more sustainable.

3.4 Subjective Indicators

For Subjective Indicators, the scores are as follows (in descending order): Users
Feedback B18(87.7)> AI Law Legislation Awareness B13(81.1)> Public Atti-
tudes B16(79.9)> Legal Regulations B17(78.3)> Policy Awareness B14(77.9)>
Forum Awareness B15(75.3)> Software Operations(74.3)> Policy Implementation
B7(73.5) > Smart Justice B9(71. 9)> Privacy Protection B10(69.2).
The average score of Subjective Indicators is 76.9.
Comparatively speaking, the average score of the objective indicators (8 items)
is 88.1, 11.2 higher than that of the subjective indicators. It shows that although
respondents have high expectations for the application of AI in the rule of law, the
conflict between the bright prospect of strong AI products and a relatively weak AI
application in reality is inevitable. This may result in the low level of satisfaction
among respondents, which forms a sharp contrast with the high score of objective
indicators. That is what we should strive to avoid in the development of AI and the
rule of law in the future.
Excerpts from the Rule of Law Forum
of the World Artificial Intelligence Conference
2019

From August 29 to 31, 2019, World Artificial Intelligence Conference 2019 was
held in Shanghai. On August 30, the Rule of Law Forum of the World Artificial
Intelligence Conference 2019, sponsored by the Organizing Committee of the World
Artificial Intelligence Conference and hosted by Shanghai Law Society and China
Judicial Big Data Research Institute, was held in Shanghai Exhibition Center. With
the theme of “Building the future rule of law and sharing the benefits of intelligence”,
the forum was attended by more than 800 legal experts and scholars, AI professionals
and technicians from all over the country and 23 countries, as well as people from
all walks of life. Yin Hong, then Deputy Secretary of Shanghai Municipal Party
Committee and Secretary of Shanghai Municipal Political and Legal Committee;
Zhang Shuyuan, Vice President of the Supreme People’s Court; and Zhang Sujun,
Vice President of China Law Society, attended the opening ceremony and deliv-
ered speeches. Gong Daoan, Vice Mayor and Director of Shanghai Public Security
Bureau; Liu Xiaoyun, President of Shanghai Higher People’s Court; Zhang Bencai,
Procurator General of Shanghai People’s Procuratorate; Ma Li, Chairman of the
China Internet Development Foundation and former Deputy Editor-in-Chief of the
People’s Daily, former Deputy Editor-in-Chief of the society; Cui Yadong, Presi-
dent of Shanghai Law Society; Wang Haiping, former President of Sichuan Higher
People’s Court; and other leaders attended the forum. Eleven experts and scholars
from China, the United States, Britain, the Netherlands, and other countries delivered
wonderful and high-quality speeches on the theme of the forum. Through in-depth
exchanges and discussions, they reached a high consensus on promoting the devel-
opment of AI and rule of law and promoting the global governance of AI and formed
a number of forward-looking, strategic, and guiding viewpoints.
Excerpt 1: It needs the joint efforts of the whole society to build the AI and rule
of law system, create the AI and rule of law ecology, and make it better benefit
mankind. Thomas believes that AI has profoundly changed our life, and everyone
is turning to build a new AI management framework and build an AI and rule of law
system. Cui Yadong believes that AI is a “double-edged sword”. While promoting the
development and application of AI, it has become an important concern of society

xlv
xlvi Excerpts from the Rule of Law Forum of the World Artificial …

to deal with the problems of AI in law, security, employment, ethics and social
governance, and to effectively prevent risks and challenges. We should be forward-
thinking and carry out research on the risks and challenges that AI may bring and
look for solutions. To build an AI and rule of law system, create an AI and rule of law
ecology, establish a regulatory system, policy system, code of ethics and morality
that is compatible with AI, and promote AI legislation, law enforcement, justice,
law-abiding and governance to avoid harm and benefit humanity requires the joint
efforts of the whole society.
Excerpt 2: The rule of law for AI is a new concept that comes into being with
the development of AI, which guarantees its development. Cui Yadong believes
that AI and rule of law are new concepts and new ideas that guarantee the healthy
and sustainable development of AI. Zhou Hanhua believes that the development of
AI brings unprecedented opportunities and challenges, and an effective legal system
can ensure its sustainable development.
Excerpt 3: AI jurisprudence is a new discipline, and it is a new and important
duty of legal professionals to promote and lead the research of AI jurisprudence.
Cui Yadong believes that AI jurisprudence is a new discipline. Faced with the risks
and challenges that AI may bring, it is one of the important responsibilities of legal
professionals and experts to promote interdisciplinary integration of AI legal research
and strive to promote AI legal research and practice.
Excerpt 4: Preventing and responding to risks and challenges should be included
in the AI national strategy. Cui Yadong believes that the current research on the
risks, challenges, and countermeasures is lagging behind the research and devel-
opment and application of AI technology, and there are problems such as lack of
systematicness and authority, and lack of unified planning and guidance in research
direction, research focus, and application of research results. We should be forward-
thinking and formulate AI national strategies to keep up with the development of AI
technology.
Excerpt 5: Artificial intelligence is a disruptive technology leading a new round
of technological revolution and industrial change, and while it creates great
benefits for mankind, it can also give rise to all kinds of new security risks. Ji
Weidong believes that the Internet of Things, AI, and big data interact to form a smart
network, and the network structure makes the role of various factors more complex
and difficult to predict. In the case of deep learning, the control and governance of AI
has become a very prominent problem. According to Shen Weixing, AI governance
risks can be divided into three levels. One is the traditional risk, that is, the loss of
privacy protection triggered by the full depth of application of the new generation of
information technology, the second is the competition between humans and machines
brought about by machine learning based on autonomous and deep learning, and
the third is a reflection on philosophy or ethics. Thomas believes that there are
controversies about personal privacy and data security in the use of AI technology.
Excerpt 6: AI requires international governance and an international consensus.
China should clarify the principles, policies, ethical standards, and guidelines
Excerpts from the Rule of Law Forum of the World Artificial … xlvii

for law-related AI research and development as soon as possible; unify specifica-

tions and forms; and strengthen international dialog. Shen Weixing believes that
it is necessary to formulate more specific principles based on current principles and
reach an international consensus. It is necessary to form an international convention
on AI governance based on principles. On the basis of the international conven-
tion, we should set up enforcement agencies and transform them into domestic rules
through the implementation of international conventions and international organi-
zations. According to Ji Weidong, the Development Plan for a New Generation of
Artificial Intelligence (2017) issued by the State Council puts forward a Chinese-style
institutional arrangement for AI development as well as nine main principles, which
are in line with the basic consensus already formed by the international community.
However, when different value orientations conflict, how to decide the meta-rules
and priority needs to be further clarified. China and other countries should clarify the
principles, policies, ethics, standards, and guidelines related to the research and devel-
opment of AI as soon as possible. Mark Beer believes that as AI technology becomes
more intelligent, most international trade will complete transactions through smart
contracts by 2029 and predicts that China will build a global blockchain platform.
Excerpt 7: Seize the opportunity, actively promote the deep integration and
application of AI and the rule of law, and better safeguard social fairness and
justice. Cui Yadong believes that it is necessary to firmly grasp the great historical
opportunity of AI and actively promote the deep integration of AI and the rule of
law. AI empowers justice, promotes judicial reform, and guarantees judicial justice.
The Shanghai AI-Assisted Criminal Case Handling System (206 System) has been
successfully developed and fully applied, effectively preventing unjust, false and
wrong cases, ensuring fair justice and enhancing judicial credibility, which is a
pioneering and successful effort to integrate AI and justice. John believes that the
application of AI will further strengthen the rule of law, make the application of
law more consistent and predictable, and can be used to eliminate prejudice. Robert
believes that the value of the application of AI technology mainly lies in improving
case management and optimizing systems, providing effective and accurate infor-
mation for judges to decide cases, and promoting rational selection and weighting of
data. Xu Jianfeng believes that efforts should be made to promote intelligent inno-
vation and reform in many aspects, such as system design and integration, to fully
realize the smart court.
Excerpt 8: Actively embrace new technologies and use new technologies such as
AI to upgrade the modernization of the national governance system and gover-
nance capacity. Cui Yadong believes that science and technology are the driving
force to promote reform and a useful tool to solve difficulties. We should actively
embrace new technologies, use AI and other new technologies to solve problems in
social governance, and upgrade the modernization of the national governance system
and governance capacity. John believes that the government can use AI technology to
actively promote legal supervision and regulation in the field of public management.
Zhou Hanhua proposed that AI and rule of law should guard against major risks
with bottom-line thinking, lead sustainable development with innovative thinking,
xlviii Excerpts from the Rule of Law Forum of the World Artificial …

establish the authority of rule of law with rule of law thinking, and promote system-
atic solutions with cogovernance thinking. At present, the most important thing in
the establishment of rule of law system is to ensure the pursuit of social fairness
and justice, the value of the rule of law, is followed through the whole process and
all aspects of legislation, law enforcement, justice and law-abiding, through concept
development, system design, and capacity enhancement.
Excerpt 9: In the era of weak AI, AI products do not have independent legal
subject qualification, so it is necessary to clarify the positioning of AI prod-
ucts and the corresponding subject of fault liability. Cui Yadong believes that the
current application of AI in the judicial field should be “assistance to case handling”
(e.g., Shanghai AI-Assisted Criminal Case Handling System). Because judicial activ-
ities have their own laws and characteristics, judges, prosecutors, and investigators
are the main bodies in handling cases. At the same time, the development of AI is
still in the primary stage, that is, the era of weak AI, which is uncertain and limited.
Therefore, at present, the application of AI in the judicial field can only assist in
handling cases, and its positioning is AI judge assistant, AI prosecutor assistant, and
AI investigation assistant, which cannot replace judges, prosecutors, and investiga-
tors in handling cases. Cruise believes that the core of AI regulation lies in how to
solve the legal personality of the AI system and the ownership attribution of the data it
generates. Thomas believes that trustworthy AI design must reflect important values
and moral principles to maintain fairness, compatibility, transparency, measurable
security, and trustworthiness.
Excerpt 10: Accelerate the training of interdisciplinary talent with legal literacy
and familiarity with AI and establish interdisciplinary disciplines of AI and law.
Cui Yadong believes that there is a shortage of talent in the field of AI and rule of
law. Those who understand law do not understand AI, while those who understand
AI do not understand law. It is urgent to adjust the subject setting, give full play to
the role of law schools, and cultivate interdisciplinary talents who understand both
AI and law. Zhang Hongwei believes that only the mastery of AI technology can
turn human knowledge into knowledge that machines can understand and use and
achieve a breakthrough. Artificial intelligence that humans can trust and control is
the only way to achieve further development.

Attachment: Introduction of Speakers

1. Cui Yadong, President of Shanghai Law Society, former president of Shanghai

Higher People’s Court, and Justice (Second Rank).
2. Topic of speech: AI-assisted Trials: Let Justice be Realized in a Visible Way.
3. Robert Maurice Brutinel, President and Chief Justice of the Supreme Court of
Arizona, USA.
4. Lecture title: AI in Arizona Courts.
Excerpts from the Rule of Law Forum of the World Artificial … xlix

5. Xu Jianfeng, Director of the Information Center of the Supreme People’s Court

and Dean of the China Judicial Big Data Research Institute.
6. Topic of speech: Accelerating the Establishment of Smart Courts with Emphasis
on AI.
7. Mark Beer, Chairman of the Metis Institute, UK.
8. Topic: The Impact of AI Contracts on International Business and Contracts.
9. Shen Weixing, Dean of Law School of Tsinghua University.
10. Topic of speech: Building the International Governance Framework of AI.
11. Thomas Cort Rubin, Honorary Lecturer, Law School of Stanford University.
12. Topic of speech: Regulatory Framework of AI: From Supervision to Governance.
13. Klaus Heine, Dean of the Graduate School of Law at Erasmus University in
Rotterdam, the Netherlands, and Chair Professor of Legal Policy Economics in
the European Union.
14. Topic of speech: The Road to Governance of AI Data.
15. Zhou Hanhua, Deputy Director and Researcher, Institute of Law of Chinese
Academy of Social Sciences.
16. Topic of speech: Good Law, Good Governance and AI Development.
17. John Frank Weaver, American lawyer, writer, doctor of law, expert in the field
of AI law.
18. Topic of speech: Witness Justice and Objectivity: AI Helps the Establishment
of the Rule of Law.
19. Zhang Hongwei, Executive Deputy General Manager of CNKI (China National
Knowledge Infrastructure).
20. Topic of speech: Knowledge Production, Dissemination and Utilization in the
Age of AI.
21. Ji Weidong, Senior Professor of Liberal Arts, Shanghai Jiaotong University,
Dean of Chinese Law and Social Research Institute.

Topic of speech: Concept, Law and Policy of AI Development.

Preface

It is an indisputable fact that the new generation of artificial intelligence has become
the new core competitiveness of future development in the world. COVID-19 swept
the world in 2020. In particular, in fighting against COVID-19 and safeguarding the
common home of mankind, the extraordinary performance of AI has given people a
better understanding of its capabilities and roles.
As a “double-edged sword”, while benefiting human beings, AI also brings risks
and challenges to human society. When paying attention to the rapid development
of AI, people are also worried about the risks and challenges it may bring.
As early as September 17, 2018, in a congratulatory letter to the World Artificial
Intelligence Conference 2018, President Xi Jinping stated, “Seizing this opportu-
nity to address the new issues raised by AI in the areas of law, safety, employ-
ment, ethics, and governmental governance requires deeper cooperation and joint
discussions among countries”.
How can we deal with the risks and challenges posed by AI?
The rule of law is the best way to regulate AI, and a sound rule of law system
is the key to the development of AI. The development of AI requires the leadership
and protection of the rule of law.
Legal experts and practitioners in Shanghai have put forward the idea of using
legal thinking and legal methods to guide, standardize, and guarantee the safe, reli-
able, controllable, and healthy development of AI. On September 20, 2018, at the
“Artificial Intelligence and the Rule of Law” High-level Seminar of the World Arti-
ficial Intelligence Conference 2018, the Shanghai Initiative on Artificial Intelligence
and the Future of the Rule of Law were released, proposing 14 initiatives on AI and
the future development of the rule of law. On August 30, 2019, “the Blue Book on
AI and Rule of Law in the World (2019)” (hereinafter referred to as “the Blue Book
(2019)”) was released at the Rule of Law Forum of the 2019 World Artificial Intelli-
gence Conference, which for the first time comprehensively collected, summarized,
and analyzed the development of AI and rule of law in China and other parts of the
world and provided valuable information and high-quality analysis reports conduc-
tive to AI legal research, leading to the development of the AI and rule of law system
and promoting the global governance of AI. The release of the Blue Book (2019)

li
lii Preface

immediately attracted great attention from all walks of life. China Central Television
and other media provided detailed reports. On August 31, 2019, at the closing cere-
mony of the 2019 World Artificial Intelligence Conference, the Congress released
the Action Plan for Building a First-Class Innovation Ecology for AI in Shanghai
(hereinafter referred to as the Action Plan for AI in Shanghai), clearly listing the
Blue Book on AI and Rule of Law as one of the “seven actions” to build a first-class
innovation ecology for AI in Shanghai. The inclusion of the Blue Book in Shanghai’s
AI development strategy is a high recognition of the research results and further
strengthens our confidence and determination to complete the Blue Book with high
quality to play an active role in ensuring the safe, reliable, controllable, and healthy
development of AI.
According to the Action Plan for AI in Shanghai and the work plan of the editorial
board of the Blue Book, the preparation of the Blue Book on AI and Rule of Law in the
World (2020) (hereinafter referred to as the Blue Book (2020)) started in December
2019. With the joint efforts of all members of the editorial board, it took 7 months
to complete the draft in June 2020. The Blue Book (2020) focuses on the latest
developments of AI and rule of law in the world from July 2019 to June 2020;
collects and summarizes the latest information on relevant strategies, policies, laws,
theoretical research, major projects and applications, and typical AI-related cases and
rule of law development in various countries; and forms a more objective, impartial,
accurate, and comprehensive report through analysis and evaluation. The Blue Book
(2020) consists of a preface, an overview of the development of AI rule of law, an
index of AI rule of law development, an assessment report on the development of AI
rule of law in China 2019, a summary of the views of the Rule of Law Forum of the
World AI Conference 2019, and an analysis report (four parts: Part I: Policy and Law
on AI, Part II: Theoretical Research on AI Rule of Law, and Part III: Application of
AI in Rule of Law, Part IV: Analysis of AI-related Judicial Cases in China), Special
Reports (six in total: Research on the Protection of Intellectual Property Rights in
AI, Research on Data Security Issues in AI Applications, Research on Legal Issues of
Facial Recognition and Privacy Protection, Research on Legal Issues Related to AI
Applications in the Medical Field, Research on Legal Issues Related to Blockchain
Technology, An Empirical Study on the Cultivation Interdisciplinary Talents of AI
and Rule of Law), “Response of AI and the Rule of Law to the Major Outbreak
of COVID-19” Special Column (“Law + AI in the Management of the COVID-
19 outbreak, the Application of AI in the Fight Against COVID-19, Application
and Legal Adjustment of Health Code in the Prevention and Control of COVID-
19”, Application and Legal Supervision of Network Direct Reporting System in the
Prevention and Control of COVID-19, Development of Epidemic Accountability
Information Disclosure Mechanism, “Prospect of Judicial Big Data Construction to
Ensure the Resumption of Work and Production by Enterprises in the Post-Epidemic
Era”. The survey is composed of the following three sections: a survey (with a focus
on prosecutorial perspectives), interviews with experts (8), a questionnaire (with
data analysis), and seven appendices. Compared with the Blue Book (2019), the Blue
Book (2020) has major changes and additions in its structure and content and has the
following new features. First, it has formulated the AI and Rule of Law Development
Preface liii

Index (the world’s first of its kind, hereinafter referred to as the Index), which mainly
includes the AI development index system, index weighting, and evaluation methods.
According to the Index, we evaluated the development of AI and rule of law in China
in 2019 and formed the Assessment Report on AI and Rule of Law Development in
2019 in China to promote the construction of AI and rule of law system. Second,
it focuses on major social events. A special chapter is dedicated to “AI and Rule
of Law Responses to Major Epidemics”, which is mainly about the role of modern
technology such as big data, artificial intelligence, and blockchain in providing rule
of law services and guarantees for the scientific prevention and control of major
epidemics. Third, its content has been greatly expanded, adding “Special Report
on the Development of AI and the Rule of Law”, “Interviews with Experts on the
Development of AI and the Rule of Law”, “Questionnaire on the Status of AI and the
Rule of Law”, and so on. Being more well organized and richer in content, the Blue
Book (2020) provides more academic and regulatory references for research on AI
and the rule of law.
In December 2019, after the editorial work of the Blue Book (2020) was launched,
the editorial department held seven special meetings to study the preparation plan,
content, and related topics (such as the AI rule of law development index, special
reports, expert interviews, questionnaires, etc.), as well as to solicit in writing from
authoritative experts in academia, well-known scholars, and members of the editorial
board, including Zhang Wenxian, Ye Qing, Liu Xiaohong, Wang Tao, Shi Weidong,
Liu Xueyao, Liu Xianquan, Peng Chenghen, Zhao Chunxue, Yang Yin, Bai Ning,
Sun Li, Yang Fei, Chen Shusen, Zhou Xiao, Hao Zuocheng, Yuan Diabao, Feng
Benben, Wang Li, Dong Liwu, and Cao Peilei, who strongly supported the writing
of this book and put forward many valuable opinions and suggestions, contributing
to the compilation of this book. This book has also received strong support from
Zhongchi Group Co., Ltd., Shanghai Hongxin Electric Power Engineering Co., Ltd.,
and Chongqing Shangyou Technology Co., Ltd. We are grateful to all of them.
The expert group of the Blue Book was responsible for researching and completing
the Index (Expert Group Leader: Cui Yadong (President of the Shanghai Law Society,
Justice of the Second Rank); Expert Group Members: Zhang Wenxian (Director
of the Academic Committee of the China Law Society, Professor, Ph.D. Super-
visor), Ye Qing (President of the East China University of Political Science and
Law, Professor, Ph.D. Supervisor), Liu Xiaohong (President of Shanghai University
of Political Science and Law, Professor, Ph.D. Supervisor), Wang Tao (Professor,
Dean of Zhejiang Tsinghua Institute of Yangtze River Delta Studies), Shi Weidong
(Vice President of Shanghai Law Society), Liu Xianquan (Professor, Ph.D. Super-
visor of East China University of Politics and Law), Peng Chengxin (Professor, Ph.D.
Supervisor of Shanghai Jiao Tong University), Xueyao Liu (Professor of Shanghai
University), Hui Peng (Researcher of Shanghai Academy of Social Sciences, Insti-
tute of Law, Ph.D.), Zhujing Lin (Prosecutor of Shanghai People’s Procuratorate,
Ph.D.). The Assessment Report on AI and Rule of Law Development in 2019 in
China was researched and completed by the assessment team of the Expert Group
of the Blue Book (Executive Director of the Expert Group: Peng Hui; members of
the Expert Group: Cui Yadong; Zhang Wenxian, Ye Qing, Liu Xiaohong, Wang
liv Preface

Tao, Liu Xueyao, Dong Liwu (Secretary of the Shanghai Law Society), Yang Yin
(Professor of the Shanghai Institute of Politics and Law), Lin Zhujing, Zhao Yucheng
(Deputy Director of the Editorial Department of the Shanghai Education News-
paper Head Office), the postgraduate team of the Institute of Law of the Shanghai
Academy of Social Sciences, and the postgraduate team of the Shanghai Institute
of Politics and Law). The summary of this book is written by Lin Zhujing. The
analysis report (Part One to Part Four) was written by Hao Zuocheng, Zhou Xun,
Peng Hui, Chen Jidong and Wu Tao. The special reports (Part Five to Part Six) are
written by Cui Yadong, Liu Xiaohong, Wang Qian, Zhao Chunxue, Yang Hua, Cao
Hongxing, Wang Chunhui, Wu Shenkuo, Zhao Yucheng, Chen Jidong, Hu Rongxin,
Chen Shusen, Qi Yongfu, Wu Haidan, Wang Huachun, Chen Shaoling, Yuan Feng,
Wu Tao, Liu Peng, Niu bingru, Zhang Shundong, Bao Kun, Hao Yuxuan, and Gong
Sihan. Expert interviews (Part Seven) were conducted with Cui Yadong, Ye Qing, Liu
Xiaohong, Liu Qingfeng, Fan Xianqun, Klaus Heine, Zhang Xinbao, Du Qian, and
other experts and scholars to form the interview report (the special report and expert
interviews basically retain the original views of the authors and experts, and the text
was compressed due to space limitations). The questionnaire (Part Eight) was written
by Zhao Yucheng. The appendix is written by Niu Bingru, Zhang Shundong, Yang
Shengyuan, Liu Yuyan, Li Yan, and Yang Liuzixian. Hu Rongxin, Hu Peng, Zhang
Zhijun, Wang Jian, Qin Kun, Xu Aijing, Xiang Mingjing, Ma Sijie, Wang Yufan,
Gong Sihan, Wang Wenjuan, Wang Ranran, Wang Yuxuan, Hao Yuxuan, Zhao Surui,
Zhang Bingjie, and Zhao Tiantian participated in the research and proofreading of
this book.
The database construction, data collection, and management work related to this
book are supported by the expert team of Xue Mengbai from the Research Center
for Rule of Law and Social Governance of Zhejiang Tsinghua Yangtze River Delta
Research Institute. To optimize the integration and analysis of data information, the
team adopted artificial intelligence techniques, including the pytorch framework for
building artificial intelligence-legal models, Spark big data computing framework,
Hbase distributed columnar database, Hive data warehouse, HDFS distributed storage
system, Mahout data mining algorithm library, Kafka Distributed message queue and
other big data technologies, as well as ElasticSearch cluster, Scrapy crawler frame-
work and other toolsets, gathered and organized cnki, lexis, westlaw, heinonline,
wiki, Google, Baidu, Amazon, and other 6PB online data resources.
The Blue Book (2020) was originally prepared by the Shanghai Law Society
and Zhejiang Tsinghua Yangtze River Delta Research Institute. The participation
of research institutions such as East China University of Political Science and Law
and Shanghai University of Political Science and Law, as well as experts such as
President Ye Qing and President Liu Xiaohong, has enhanced the authority of the
book.
Due to inexperience and limited information (interviews and research were
affected by COVID-19), the contents of this book, including the collection, summary,
and analysis of information, as well as the text, may contain omissions or errors.
We would like to invite readers to criticize and correct them. We also expect more
suggestions for our future writing.
Preface lv

The future is here! The Blue Book of AI and Rule of Law in the World (2020) will
continue to contribute to promoting the construction of the AI rule of law system;
creating an AI rule of law environment; leading AI rule of law research; building a
hub of AI and rule of law; serving to promote AI global governance; and promoting
the safe, reliable, controllable, and healthy development of AI.

Shanghai, China Yadong Cui

July 2020
Contents

1 AI Policies and Laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Yadong Cui
2 Theoretical Research on AI and the Rule of Law . . . . . . . . . . . . . . . . . . 41
Yadong Cui
3 AI Applications in the Rule of Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Yadong Cui
4 Al-Related Case Analysis (China) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Yadong Cui
5 Special Reports on the Development of Artificial Intelligence
and the Rule of Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Yadong Cui
6 The Use of AI in Law in Response to Major Epidemics . . . . . . . . . . . . . 207
Yadong Cui
7 Expert Interviews on the Development of Artificial Intelligence
and the Rule of Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Yadong Cui
8 Questionnaire on Legal Aspects of Artificial Intelligence
(with Data Analysis) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Yadong Cui

Appendix A: List of Policies and Regulations on Artificial

Intelligence (2019.7–2020.6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
Appendix B: List of Artificial Intelligence and Law Research
Institutions (2019.7—2020.6) . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
Appendix C: List of Important Publications and Papers
on Artificial Intelligence and Law (2019.7—2020.6) . . . . . . 359

lvii
lviii Contents

Appendix D: List of Important Conferences (Forums) on Artificial

Intelligence and Law (2019.7—2020.6) . . . . . . . . . . . . . . . . . . 361
Appendix E: List of Typical AI Applications in the Judicial Field
(2019.7—2020.6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
Appendix F: List of Typical AI-Related Cases (2019.7—2020.6) . . . . . . . 365
Appendix G: Chronicle of Events of AI and Law . . . . . . . . . . . . . . . . . . . . . 367
About the Editor

Cui Yadong President of Shanghai Law Society, former

President of Shanghai High People’s Court, Justice
(Second Rank) of the People’s Republic of China.
He joined the Chinese People’s Liberation Army
(PLA) in December 1969 and the police in January
1974. As a soldier, he served in the Signal Company
of the 74 Division of the PLA. Later, he was trans-
ferred to the Hefei Municipal Public Security Bureau.
Starting from an ordinary police officer, he served as
the Deputy Section Chief of the Political Division, the
Deputy Director of the Case Preinvestigation Division,
the Director of the Detain for Interrogation Center along
the way before being promoted as the Deputy Director,
and later the Director of Hefei Municipal Public Secu-
rity Bureau. After that, he was appointed succes-
sively as the Deputy Director-General and the Director-
General of Anhui Provincial Public Security Depart-
ment; the First Political Commissar of Anhui Provin-
cial Corps of Chinese People’s Armed Police Forces.
Then, he was assigned to Guizhou Province, holding
different positions such as member of the Standing
Committee of the CPC Guizhou Provincial Committee,
the Head of Subcommission for Political and Legal
Affairs under CPC Guizhou Provincial Committee, the
Director-General and the Deputy Police Supervisor of
Guizhou Provincial Public Security Department, the
Director of Social Stability Maintenance Office of
Guizhou Province, the First Political Commissar of
Guizhou Provincial Corps of Chinese People’s Armed
Police Forces, and the President of Guizhou Provincial
Law Society. In April 2013, he served as the Acting

lix
lx About the Editor

President of Shanghai High People’s Court, and in

January 2014, he served as the President of Shanghai
High People’s Court and Justice (Second Rank) of the
People’s Republic of China. He became the President of
Shanghai Law Society in October 2018.
He is the author of monographs such as Emer-
gency Management of Mass Incidents and Social Gover-
nance (published by Press of Party School of the
CPC Central Committee), A Rule of Law Country (by
People’s Publishing House), Artificial Intelligence and
Judicial Modernization (Chinese edition by Shanghai
People’s Publishing House; English edition by Springer-
Verlag, Germany). He is also the Editor-in-Chief of
the books including Exploration and Practice of the
Reform of the Judicial System of Courts in Shanghai
(by People’s Court Press), and Sunshine Illuminates the
Path of Return (by Press of China National School of
Administration).
Chapter 1
AI Policies and Laws

Yadong Cui

1 Introduction

“AI policies” in this part refers to AI development strategies, implementation plans

and other policy documents issued by central and local governments and government
agencies in all countries, as well as policy recommendations proposed by interna-
tional organizations. “AI laws” herein refer to regulatory documents of universally
binding force on AI released by legislatures, government agencies, local governments
and international organizations.
“Policies and laws” hereof is a combined name of “policies” and “laws” involving
AI. “Relevant policies and laws” herein refer to “policies and laws”, with major
content not on AI yet implicating it. “Specific policies and laws” herein refer to
“policies and laws” that focus on AI.
This section will analyze the development and trend of AI policies and laws in
China and beyond (some countries and regions in the world) from July 2019 to June
2020. The overview comes as follows.
First, the basic situation in China. AI policies and laws have developed with a
step-by-step transition from macro planning to specific implementation. China has
carried out specialized implementation in areas such as education, the environment,
intelligent vehicles, and improved safeguard measures in an innovative manner by
governance principles, innovation platforms, and innovative development pilot zones.
From the level of the nation and departments of the State Council, documents of AI
programs during this period focus on three fields of industry, scientific research
and education. In addition, relatively great attention has been given to government
services, institutions, social governance, infrastructure, pilot zones, and epidemic
response. Meanwhile, AI-involved documents by local governments have centered

Y. Cui (B)
Shanghai Law Society, Shanghai, China
e-mail: [email protected]

© Shanghai People’s Publishing House 2022 1

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2_1
2 Y. Cui

on industry and scientific research. Shanghai, Beijing and Anhui provinces have
made efforts mainly in ecology innovation, data opening, and industry support.
Second, the basic situation beyond China. Policy development in major coun-
tries and regions in the world is similar to that in China, already going from the
macro level gradually to specific measures of specialized implementation and safe-
guard measures. The US, EU and other countries and regions have introduced
concrete measures targeted at many aspects, including fundamental R&D invest-
ments, government resource opening, innovation barrier removal, international envi-
ronment creation, governance principle exploration, and advancement in internal
cooperation, education training, and public–private cooperation.
At the legal level, major countries and regions in the world have gradually recog-
nized the potential social problems appearing in the process of AI development
and provided the framework and recommendations for principles on AI gover-
nance. Nevertheless, given the possibility of barriers to prevent innovation, they
have passed few rules and regulations to impose mandatory restrictions. The US
takes a cautious attitude toward the adoption of rules and regulations, while the EU
is relatively more active by contrast. Their AI governance principles currently focus
on social welfare, fairness, collaboration, good governance, transparency, safety,
privacy, accountability, data quality, and human autonomy, the major point of each
aspect being different based on their own national situations.

2 AI Policies and Laws in China

2.1 The Basic Situation

Since the State Council of China released the New Generation Artificial Intelli-
gence Development Plan in 2017, the AI-specific policy system in China consists
of three parts: core strategies, specialized implementation and safeguard measures.
Of them, core strategies mainly include the New Generation Artificial Development
Plan by the State Council in July 2017 and the Guidelines for the Deep Integration
of Artificial Intelligence and Real Economy by the Central Comprehensively Deep-
ening Reforms Commission in March 2019. Under the guidance of the above two
documents, specialized documents of safeguard measures include governance princi-
ples, innovation platforms and innovative development pilot areas, while specialized
implementation mainly contains education, environment, intelligent cars and other
specific directions.
According to noncomprehensive statistics, this part compiles a total of 122 policies
and laws relevant to AI from July 2019 to June 2020, with 28 at the national level
and 94 departments of the State Council.
Generally, current AI development has gradually transitioned from macro
planning to specific measures of specialized implementation and safeguard measures.
1 AI Policies and Laws 3

2.2 National Policies and Laws

National policies and laws herein are introduced by entities such as the Central
Committee of Communist Party of China (CPC), the Central Comprehensively Deep-
ening Reforms Commission, the National People’s Congress, the State Council, and
general offices of the CPC Central Committee and the State Council. Their contents
mainly focus on AI-related policies and laws.
According to noncomprehensive statistics, there was no release of legal docu-
ments targeting AI at the national level from July 2019 to June 2020. Rather, the
mention of AI is scattered in other relevant policies and laws. Twenty-eight national-
level policies and laws with AI involved have been introduced, where the focus
in AI-related projects goes to industry, scientific research, education, government
services, institutions, and social governance with an involvement of infrastructure,
epidemic response, safety, taxation, pilot zones, credit supervision and management,
finance, agriculture, medical treatment, employment, trade, transportation, engi-
neering, market regulation, manufacturing, cultural tourism, sports, national defense,
international cooperation, and ethics.
From the frequency of various items mentioned in overall policies and laws at the
national level, particular attention is given to industry, scientific research, education,
government services, institutions and social governance. This section will mainly
introduce the first three dimensions.

2.2.1 AI Industry

Documents of this dimension mainly involve policy orientation of the AI industry,

measures facilitating the development of the AI industry and the role of the AI
industry in society. More details are as following:
The first is industry policy with the basic orientation of fostering leading enter-
prises and future industry. The Outline for Integrated Regional Development of
the Yangtze River Delta by the CPC Central Committee and the State Council
in December 2019 pointed out, “the science and technology innovation is tightly
integrated with industry… AI and other new technologies infiltrate and integrate
with the traditional industry…we should focus on…AI…in ten key areas…and
nurture leading companies of international competitiveness. With the orientation
toward…the next generation AI…in eight key areas, the step should be quickened
to develop the future industry layout.” With the emphasis in the industry policy
on fostering leading companies, it aims to gather power from these companies to
enhance internal competitiveness, while with orientation toward the future industry
layout, it aims to seize new development opportunities in the future industry.
The second is the promotion methods to obtain the AI industry delivered by pilot
areas, demonstration zones, international cooperation and military-civilian integra-
tion. According to the Overall Plan for Lingang New Area of China (Shanghai)
Pilot Free Trade Zone unveiled by the State Council in July, 2017, it is proposed to
4 Y. Cui

“build AI innovation and application demonstration zones, speed up the opening of

application scenarios, and promote development of intelligent cars, manufacturing
and robots and other new industries and business forms.” In August, 2019, the State
Council released the Overall Plan for Six New Pilot Free Trade Zones, saying “the
Jinan pilot zone will focus on the development of AI… and other industries, conduct
comprehensive pilot trials of the new open economy system…enhance the industry
cooperation with the ASEAN member countries in the Beidou Navigation Satel-
lite System (BDS), big data, AI and other industries…accelerate the building of the
5G industry system, promote the deep integration of internet, big data, and AI with
the real economy…as well as the industry system of national defense with the local
industry system, and actively develop new industries and key areas in military-civilian
integration, including aerospace, ocean engineering equipment, new materials, new
energy and AI.” In these documents, pilot areas, demonstration zones, international
cooperation and military-civilian integration are four major measures specialized to
help get the AI industry delivered and applied during this period.
The third is AI industry investment and layout to leverage social functions, such
as employment promotion. In December 2019, the Opinions of the State Council on
Further Stabilizing Employment clearly stated, “the strategic development project
of emerging industrial clusters should be further promoted and the infrastructure
investment and the industry layout in AI, industrial internet and other areas should
be enhanced.” Due to the COVID-19 impact, “stabilizing employment” has been
upgraded as the current priority on the national work agenda. Investment and layout
for the AI industry will help ease employment pressure in the short term, while in
the long run, apart from helping the AI industry experience the boom, it can play an
active role in society with an emphasis on improving employment quality.

2.2.2 Scientific Research

Documents herein mainly involve ethics of AI scientific research and the function
of AI scientific research in the development of local areas and particular industries.
Specifically, China improves national AI scientific research ability by organizing
and building the National Science and Technology Ethics Committee and scien-
tific research centers, encouraging leading companies to establish innovation plat-
forms, reducing taxes on R&D companies and promoting the integration of scientific
research and the industry.
Documents herein mainly involve ethics of AI scientific research and the function
of AI scientific research in the development of local areas and particular industries.
Specifically, China improves national AI scientific research ability by organizing
and building the National Science and Technology Ethics Committee and scientific
research centers, encouraging leading companies to establish innovation platforms,
reducing taxes on company R&D and promoting the integration of scientific research
and the industry.
The first is to build the National Science and Technology Ethics Committee for
guiding AI scientific research to move toward the right direction. In July 2019,
1 AI Policies and Laws 5

the ninth meeting of the Central Committee for Deepening Overall Reform in an
All-Around Way approved the plan to establish a national committee for ethics in
science and technology, saying “scientific integrity and professional ethics are the
common values that scientific and technological activities have to abide by. The
establishment of the national committee aiming to enhance overall regulation and
guidance and coordination will promote the development of a more comprehensive,
oriented, ordered and coordinated governance system for science and technology.
Immediate actions should be taken to improve institutional regulations, strengthen
the governance mechanism, intensify ethical supervision and detail relevant rules
and regulations as well as those for ethical checks to regulate all scientific and
research activities.” With continuous breakthroughs in basic AI technologies and
ever widening application scenarios, the clash between ethics and AI technology
tends to be increasingly fierce. What the Science and Technology Committee can
help is to set ethical principles the AI technology is subject to in development.
The second is to build research centers, encourage leading companies to establish
innovation platforms and reduce taxation on R&D companies for advancement in
local AI scientific research development. In December 2019, the Outline by the
CPC Central Committee and the State Council proposed that “efforts should be
done to enhance…AI…and other new technologies in R&D and applications, and
to support major companies in joining hands with research institutes to build the
AI R&D platform covering the Yangtze River Delta and other new ones alike…”
Earlier in August 2019, the CPC Central Committee and the State Council issued
the Opinions on Supporting Shenzhen in Building a Pioneering Demonstration Zone
for Socialism with Chinese Characteristics, saying “Shenzhen should be supported
in building 5G, AI, cyberspace in science…and other major innovation carriers and
in exploring to build an international technology hub…” In July 2019, the State
Council in the published Plan for Lingang New Area in science…and other major
innovation carriers and in exploring to build an international technology information
hub…” In July 2019, the State Council in the published Plan for Lingang New Area
proposed at 15 percent of the industry, this period, companies engine at the next five
years engine to conduct in the industry. By supporting leading companies to build
R&D centers together with tax-cut policies, it helps encourage enterprises to make
investments in fundamental R&D, and by supporting leading companies to build
innovation platforms, it helps develop the AI ecological chain of medium-, small-
and microsized (MSM) enterprises. The foresaid measures will be specific methods
available for local areas to develop AI scientific research.
The third is to promote deep integration of AI scientific research and the industry
to promote development in transportation and other specific industries. In the Outline
for Building China’s Strength in Transport released by the CPC Central Committee
and the State Council in September 2019, it recommends to “align with the new gener-
ation information technology, AI, intelligent manufacturing…and other cutting-edge
technologies in the world, to strengthen forward-looking and disruptive technology
research that can trigger transformation in the transport industry, and to promote
the deep integration of big data, internet, AI, the blockchain technology, super-
computers and other new technologies with the transport industry.” In the AI field,
6 Y. Cui

scientific research is required to solidify the foundation and effective applications to

provide sufficient intrinsic motivation for scientific research. The deep integration of
“industry, education and research” will be an effective way to develop AI in specific
industries.

2.2.3 Education

Documents hereof are to capitalize on AI in education mainly through three priorities:

developing new teaching methods and an individualized education system through
AI, strengthening AI applications in ideological and political education and using AI
to empower education management to help relieve the burden on high and primary
school teachers.
The first is to develop new teaching methods and establish an individualized educa-
tion system through AI. Adopted in October 2019, the CPC Central Committee’s
decision on some major issues concerning how to uphold and improve the system
of socialism with Chinese characteristics and advance the modernization of China’s
system and capacity for governance clearly states, “it should tap into the edges of
internet education and AI for innovating education and learning methods, speed
up the building of a more flexible education system available for and suitable to
everyone, to build a learning society.” China has paid full attention to the revo-
lutionary transformation AI can bring to individualized education. Transformation
in the education philosophy and learning methods can further cement positions in
advantageous education areas and gain impetus for innovation in society.
The second is to strengthen AI applications in ideological and political educa-
tion. In August 2019, the general offices of the CPC Central Committee and the
State Council published some opinions on Deepening the Reform and Innovation of
School Ideological and Political Theory Course in the New Era, clearly proposing
to “promote applications of AI and other modern information technologies in ideo-
logical and political education for building a batch of national-level virtual imitated
teaching and experience centers for ideological and political courses.” Integrating
AI and other modern technologies with ideological and political education is a key
direction to innovate reforms in the future national ideological and political theory
course.
The third is to use AI to empower education administration and reduce the burden
on teachers in high and primary schools. In December 2019, the general offices of
the CPC Central Committee and the State Council in Some Opinions on Relieving
Burden on Teachers in High and Primary Schools for Further Creating Favorable
Environment for Education and Teaching recommended “a full uptake of modern
information technologies, in particular the AI technology, for enhancing informati-
zation and scientificity in educational administration substantially to have teachers
running fewer errands by more information sharing.” The work overload of high
1 AI Policies and Laws 7

and primary school teachers aroused government attention. AI-empowered educa-

tion administration can facilitate raising basic education levels by enhancing admin-
istrative efficiency, laying the sturdy foundation for national talent training in the
future.
For AI development, scientific research is a prerequisite, industry is the carrier and
education is the foundation. With great importance attached to scientific research,
the industry and education in the overall documents at the national level, it fully
represents the characteristic of China to highlight the essential role of foundation in
the AI cause and introduce policies from both long-term and short-term perspectives.

2.3 Policies and Regulations by Departments

Policies and regulations herein are mainly published by the departments of the State
Council, and their major contents are about specialized policies and regulations
relevant to AI.
There are 94 AI-relevant documents of policies and regulations by the departments
of the State Council, with only a fraction of specific policies and regulations for AI
and the majority of them relevant to AI. In addition to 28 overall documents at the
national level, the total amounts to 122 documents.
According to the statistics of the frequency that AI is mentioned in different
areas in policies and regulations at the national and department levels in China,
the key areas remain on industry, scientific research and education (each mentioned
44 times, 33 times and 19 times, respectively, one area more than once mentioned
in the same document calculated only once). China put relatively much attention
on areas such as infrastructure, systems, pilot zones, government services, epidemic
response, ethics, finance, transportation and pension (7–12 times). Meanwhile, it also
involves manufacturing, environment, medical treatment, social influence, safety,
employment, social governance, journalism and communication, agriculture, cultural
tourism, talent introduction, taxation, enterprise internal control, trade, credit super-
vision and management, engineering, market regulation, sports, national defense,
international cooperation, intellectual property and postal services (1 to 6 times).
At this time period, the specialized AI policies and legal documents by depart-
ments of the State Council stress pilot zones, open innovation platforms and educa-
tion, generally speaking, in compliance with documents at the national and depart-
ment levels where the six most mentioned aspects are (industry, scientific research,
education, infrastructure, systems and pilot zones). This section will elaborate on AI
pilot zones, open innovation platforms and education, the current key areas in China.

2.3.1 Pilot Zones

Pilot zones, the short form for the national new generation AI innovation and devel-
opment pilot zones, refer to areas that take a step ahead in trials with leadership
8 Y. Cui

and motivation to promote AI innovative development by AI technology demonstra-

tions, policy trials and social experiments in local areas. According to Guidelines
for National New Generation Artificial Intelligence Innovation and Development
Pilot Zone Construction Work by the Ministry of Science and Technology (MoST)
in August 2019, the AI innovation pilot zone can be explained from construction
principles, construction goals and overall layout and key tasks.
The first is to follow the development principles of application-driven local
stakeholders, pioneering policies and unique characteristics. Application-driven
approaches promote the application of AI in the economic and social fields and
the iterative upgrade of AI technologies and systems through large-scale applica-
tions. In terms of local stakeholders, full play is given to the mainstay role of local
institutions in the construction of pilot zones to make bold explorations in promoting
the development and application of AI. Pioneering policy refers to a pioneering role
in the development of AI systems and mechanisms, policies and laws, as well as regu-
lations and standards for a more complete policy system. What unique characteristics
mean is to combine local economic and social conditions with the basic conditions
for AI development to form unique experiences, practices, and development models
in the deep integration of AI and economic and social development and governance
in the intelligence age.
The second is to set up approximately 20 pilot zones by 2023 and complete
pioneering goals and layouts in regions with rich AI innovation resources and strong
development foundations. As of 2023, approximately 20 pilot zones will be estab-
lished as a group of AI innovation hubs to explore new policy tools and typical
models and accumulate replicable experience. An important layout of unique char-
acteristics will be conducted and centered on regions with relatively mature condi-
tions for AI development, including the Beijing-Tianjin-Hebei region, the Yangtze
River Economic Belt, the Guangdong-HongKong-Macao Greater Bay Area, and the
Yangtze River Delta.
The third is to complete key tasks of exploring the deep integration path of AI
with social and economic development, creating a favorable institutional environ-
ment, conducting social experiments and constructing infrastructure. The deep inte-
gration task calls for carrying out demonstrations of AI technology applications and
exploring the new path in support of deep integration of AI with social and economic
development. The task of creating an institutional environment conducive to AI inno-
vation and development requires forming a policy framework and a system of regula-
tions and standards adapted to AI development in data openness and protection, the
application of research achievements, intellectual property, security management,
talent introduction, finance and taxation, social security, and international cooper-
ation. The task of social experiments is to keep objective records and assessments
of the comprehensive impact of AI technology on the behavior of individuals and
organizations, employment structures, and changes in income to continuously accu-
mulate data and practical experience in support of governance in the intelligence age.
The construction task of AI infrastructure is to strengthen the construction of infras-
tructure in networks, big data, and computing, AI R&D bases and open innovation
platforms to promote the secure and orderly release of public data.
1 AI Policies and Laws 9

In accordance with the spirit of the Guidelines for Pilot Zone Construction Work,
in October and November 2019 and January 2020, MoST made replies to support
Tianjin, Shenzhen, Hangzhou, Hefei, Deqing County, Chengdu, Xi’an and Jinan
to take a lead in setting up pilot zones in combination with their own advantages
and characteristics. Each pilot zone is set with different tasks and goals. MoST sent
letters to support Beijing in February 2019 and Shanghai in May 2019 in building the
national new generation AI innovation and development pilot zone. Together with
the above eight pilot zones, at least ten AI innovation pilot zones have been created
in China.

2.3.2 Open Innovation Platforms

Open innovation platforms, the abbreviated form for open innovation platforms for
next-generation AI, are an important innovation vehicle with a focus on key segmen-
tation to continuously export core R&D and service capabilities of AI by completely
leveraging the leadership and demonstration role of leading enterprises and research
institutes in the AI industry and effectively integrating resources of technology, indus-
trial chains and finance. According to the Guidelines for Pilot Zone Construction
Work by MoST in August 2019, the open innovation platform policy emphasized
construction goals, construction principles and key tasks.
The first is to achieve the construction goals, including encouraging technology
opening and sharing, increasing R&D investments in the entire society, and advancing
the promotion and application of AI technology achievements. Establishing open
innovation platforms can encourage the opening of all universal software and tech-
nologies while supporting people in practicing innovation and entrepreneurship in
society as well as teams and MSM companies dedicated to AI technology R&D
to promote and apply AI technology achievements. China has fully recognized the
significance of R&D for AI development. The technology open source and leader-
ship of major companies contribute to forming an R&D ecological chain with MSM
companies involved in proceeding with achievement promotion and application.
The second is to follow the construction principles of application driven, leading
companies as stakeholders, the market mechanism and collaborative innovation.
What the application driven means is to be oriented toward major application needs
for AI. Stakeholders include pioneering companies in AI segmentation that build
open source platforms where R&D resources of AI technology are available to
everyone and export AI technology service ability to society in contribution to AI
technology applications in the industry. By nurturing leading companies, it can facil-
itate the development of MSM companies. The market mechanism encourages the
market-based mechanism for organizations and management with companies as the
mainstay investing in open innovation platforms. Collaborative innovation encour-
ages the joint efforts of local governments, the industry circle, scientific research
institutes and high schools to promote the building of open innovation platforms
and through the integration of talent, technology, data industrial chains and other
resources to create open ecology.
10 Y. Cui

The third is to complete key tasks of basic technology innovation, achievement

promotion and applications, open sharing service provision and guidance on MSM
companies and industry developers to practice innovation and entrepreneurship.
Basic technology innovation refers to collaboratively developing AI basic theo-
ries, model methods and fundamental research on software and hardware to serve
and support cutting-edge basic theories of AI and key technology innovation. The
promotion and application of achievements actively explore the mechanism for
conversion and applications of the achievements generated from open innovation
platforms. To provide open sharing services is to open datasets and form standard-
ized modular models, middleware and application software through open interfaces,
model libraries and algorithm packages to provide open sharing services of hard-
ware and software. The aim of guiding MSM companies and industry developers to
conduct innovation and startups is to build a community for information sharing and
experience exchanges in the segmentation, lowering the entry barrier for the use of
technologies and resources.

2.3.3 Education

To implement the key layout of the CPC Central Committee and the State Council to
accelerate the development of the new generation AI, promote “double first-class”
construction of colleges and universities with great efforts to build the AI talent
training system above the advanced level in the world, the Ministry of Education,
National Development and Reform Commission and the Ministry of Finance issued
the notice of Several Opinions on “Double First-Class” Construction of Colleges and
Universities to Promote the Integration of Disciplines and Speed up the education
of Postgraduate Students in the Field of Artificial Intelligence in January, 2020. The
notice under the principles of need orientation, application driven, project guidance,
multiple support, cross-sector integration and targeted training has established in
the education sector a great number of key measures, including scientific layout,
creating an industry-education integrated platform, strengthening the cooperation
between universities and companies, improving the disciplinary setting and evalua-
tion, enhancing the construction of the curriculum system, intensifying international
exchanges and cooperation and expanding the training scale.

2.3.4 Infrastructure

If AI development is compared to a speeding car, then scientific research and educa-

tion are its “engine”, the industry its “fuel oil”, and infrastructure its “chassis”. Well-
built infrastructure can provide strong support for AI development. The Ministry
of Industry and Information Technology issued a Circular about Consolidating the
Industrial Chain and Pushing for Coordinated Resumption of Work and Production
in March 2020, clearly saying to “accelerate the construction of the 5G network, IoT,
big data, AI, industrial internet, smart city and other infrastructure”. The National
1 AI Policies and Laws 11

Development and Reform Commission, the State Council Taiwan Affairs Office
and eight other agencies jointly released a Circular Requiring Coordinated Efforts
to Support the Development of Taiwan-Funded Enterprises and Advance Taiwan-
Funded Projects amid the Coronavirus Epidemic, also proposing “to support Taiwan-
funded enterprises…through multiple ways in participating in the R&D, production
and construction of the mainland 5G, industrial internet, AI, IoT and other new
infrastructure.”
Looking forward, plans relevant to “new types of infrastructure” with AI infras-
tructure as an important point will be carried out step by step. According to the
Report on the Implementation of the 2019 Plan for National Economic and Social
Development and on the 2020 Draft Plan for National Economic and Social Devel-
opment approved at the third session of the 13th National People’s Congress on May
2020, it proposes to “introduce policies for supporting the construction of new types
of infrastructure, and promote investment in new infrastructure including 5G, IoT,
the Internet of Vehicles, industrial internet, AI, and the national data center.” It is
estimated to roll out policies relevant to AI infrastructure any time soon.

2.3.5 Response to COVID-19

In addition to what is mentioned above, AI has been an important means for the
prevention and control of COVID-19 in documents since February 2020. In February,
the comprehensive team in the Joint Prevention and Control Mechanism of the State
Council issued a Circular on Developing Online Services for Further Enhancing
Epidemic Control and Prevention in Hubei, going “it will provide telemedicine
services and organize an AI expert group…to provide patients with free coun-
seling.” Assistance from the AI expert group is regarded as a significant method
in support of epidemic control and prevention in Hubei. From February to March
in 2020, among all the documents, including the Guidelines on Promoting Orderly
Work Resumption of Industrial Telecommunications Enterprises by MIIT, the Guide-
lines on Information Construction and Application for the Epidemic Prevention and
Control in the Community jointly by the General Office of Civil Affairs, the Secre-
tary Bureau of the office of the Central Cyberspace Affairs Commission, MIIT and
the National Health Commission, the Notice on Promoting Orderly Work Resump-
tion of Enterprises in the National High-Tech Zone by MIIT, and the Notice on
Fully Leveraging the Role of Industry Associations and Chambers of Commerce
in Supporting the Work and Production Resumption of Private Medium and Small
Companies by the general offices of the National Development and Reform Commis-
sion and Civil Affairs, it requires the AI technology to provide technical support
in resumption of work and production, epidemic analyses, virus testing, vaccine
R&D, community epidemic control, assistance for medium and small enterprises
and many other key aspects in epidemic response. During epidemic prevention and
control, China attached great importance to capitalizing on AI technology, providing
essential support and guarantees for curbing the spread of the virus.
12 Y. Cui

2.4 Local Policies and Regulations

Policies and regulations herein are mainly published by local governments in China,
as well as the Economy and Information Technology Commission, the Industry and
Information Technology Bureau and other departments. The major contents contain
specialized policies and regulations for AI.
With AI macro planning relatively mature since 2017, local areas have already
begun gradually to devise specific implementation schemes for AI planning,
presenting a trend from the provincial level to the prefectural level.
With regard to content, the policy documents issued by Anhui Province are about
innovation in industrial scientific research and industrial bases. Beijing Municipality
has released policy documents on robot innovation, data opening, park innovation,
education development and other aspects. The documents in Shanghai Municipality
are targeted at scientific research innovation platforms. In December 2019, the
Yunnan government designed AI macro planning for the province. Generally, the
essence of each policy document in local areas focused on industry and scientific
research, scientific innovation, data opening, park construction and other aspects.
According to noncomprehensive statistics, Guangzhou, Hangzhou, Chengdu,
Jinan, and Weihai cities introduced AI-related policies designed for their own AI
development during the same period of time.
This part will conduct a detailed analysis of three documents as typical examples:
the Action Plan for Building Shanghai into an AI Hub with World-Class Innova-
tion Ecosystem from 2019 to 2021 (hereafter Shanghai Action Plan) by Shanghai
Commission of Economy and Information in September 2019, the Work Plan to
Boost AI Industry Development through Opening Public Data (hereafter Beijing
Work Plan) by Beijing Big Data Promotion Group Office in October, and the Notice
of Several Policies on Supporting Innovation and Development in the AI Industry
(hereafter Anhui Notice) by the Anhui People’s Government in March 2020.

2.4.1 Shanghai Innovation Ecology Policy

The core goal of the Shanghai Action Plan focuses on “innovation”, highlighting
three key targets to incentivize the innovation vitality of all parties, strengthen orig-
inal innovation motivation and improve innovation capabilities for support. Concrete
measures are taken with efforts from multiple directions, such as an optimized layout
of innovation bases, more room reserved for new-type innovation, greater efforts to
lower innovation cost, establishment of platforms for frontier theory research, an open
source community, the industrial application and innovation, the major deployment
of computing power services, and ordered data opening and exchanges, talent gath-
ering and training, creation of world-level application scenarios, speeded formation
of capital circulation, stronger boost to market innovation, more efficient conversion
of intellectual achievements and better environment for the rule of law. In addition,
in the Action Plan for Shanghai Promoting the Construction of New Infrastructure
1 AI Policies and Laws 13

from 2020 to 2022 in May 2020, four key areas in “new infrastructure construc-
tion” with Shanghai characteristics were identified. Each of the four is centered on
building the “innovation ecology”, including the construction of the “new network,
facility, platform and terminal” with respective centers on the infrastructure of the
new generation network, innovation bases, integration of AI and other technologies
and intelligent terminals.
It is worth noting that Shanghai emphasizes not only the direct push to AI through
industry policies but also the full utilization of enterprises through market mech-
anisms. By creating platforms to lower transaction costs and by talent training
and introducing capital, achievements of governance and other factors to realize
the construction of AI ecology, a closed loop of AI development ecology can be
formed. Meanwhile, Shanghai has stressed carrying out research on information secu-
rity, privacy protection, ethics, and rules and regulations in the AI sector. Creating
AI governance in the “Shanghai mode” by actively advancing AI legislation and
releasing reports on AI in the rule of law on a regular basis will provide “infras-
tructure” and “soft power” to AI innovation ecology from the perspective of the
rule of law. In general, Shanghai has set a “sample” for AI industry development in
other areas by integrating the industry policy of tax incentives and subsidies and the
competition policy of creating platforms and markets in combination with the rule
of law.

2.4.2 Beijing Data Opening Policy

The Beijing Work Plan is a measure of great importance to take a lead in opening
public data across the nation. Just as the Chinese saying goes, “before the military
force moves, fodder and provisions go first.” If data are “fodder and provisions” for
AI development, then data opening can be called “fodder and provisions precede”
development. Through the policy of opening public data, Beijing attempts to guar-
antee society without conditioning the opening of public data focused on key areas of
finance, credit, transportation, medicine, transportation, and justice. It will construct
the innovation base of open public data where a batch of special public data is
available with conditions to local AI enterprises in application competitions and
other manners. Specific measures include data hierarchy management, data opening
with/without conditions, creation of innovation bases for open public data, appli-
cations of AI in city management and public services, and the construction of the
AI ecological system. In hierarchical management, data are categorized into four
levels: sharing with conditions, sharing without conditions, opening with conditions
and opening without conditions. The general direction proceeds mainly through
government guidance plus the leadership of core major enterprises, together with
the establishment of ecology alliances and platforms for algorithms and computing
power, as well as the enhancement of the ownership and protection of AI-involved
patents. Beijing is regarded as one of the key areas with AI enterprise clusters. Hence,
its data opening policy is of great and nationwide demonstration significance. Due to
data privacy, data ownership and many other legal problems involved in the opening
14 Y. Cui

of data, Beijing, as a practice base, will provide fresh samples and trial experience
to formulate rules and regulations for data and AI in the future.

2.4.3 Anhui Industry Support Policy

The Anhui Notice rolled out relatively specific industry innovation policy plans
roughly from ten aspects, including improvement in innovation abilities, estab-
lishment of supporting platforms, support in project construction, advancement
in demonstrations and applications, progress in data opening, speeded industry
clustering, creation of national pilot zones, increased fund support, and intensi-
fied industry services and talent support. Anhui province provides relatively large
amounts of money to support AI industry innovation: the fund to support the innova-
tion and R&D tops 5 million yuan; to fund entities responsible for building platforms
for AI public services as well as open source and generic technologies ranges from
1 million yuan to 3 million yuan; to finance intelligent sensor projects and others is
a maximum of 20 million yuan; to endow demonstration programs tops 10 million
yuan. Anhui Province also emphasizes advancing data opening, with great efforts
to assist medium and small companies. In support of SMS companies that consume
services by companies in the list of shared AI computing resources, the govern-
ment will subsidize 30% of the actual price paid, the same unit granted up to a
total of 300,000 yuan. AI industry innovation policies in Anhui Province, through
fund support, data opening, talent incentives, and industry services, have a rela-
tively comprehensive coverage of industry, education, research, and application in
AI innovative development. Hence, Anhui Province is one of the areas with relatively
detailed AI-relevant policies released at an early time in China.

3 AI Policies and Laws Beyond China

3.1 Overview

3.1.1 From Policies to Implementation Measures

In terms of AI policies, major countries and regions in the world have translated
from the deployment of macro planning in 2016 to 2017 to the implementation of
specific measures in 2019 to 2020. After rolling out the macro plan of Maintaining
American Leadership in AI in February 2019, the US Republican government made
progress in adopting specific documents in fundamental R&D investments, govern-
ment resource opening, innovation barrier elimination, government service improve-
ment, international environment improvement, and governance principle exploration.
The EU introduced macrolevel artificial intelligence for Europe in April 2018, and
then in February 2020, the White Paper—A European Approach to Excellence and
1 AI Policies and Laws 15

Trust—was published to advance AI development with specific measures to enhance

internal cooperation, R&D and innovation, education and training, and private and
public cooperation. Compared with the foregoing documents, macroplans in Russia,
the ROK, and Singapore came up at a later time. Nevertheless, once released, they
also contribute to progress in application promotion, industry policies, talent plans
and public awareness, and later concrete measures are formulated.

3.1.2 Governance Principles in Legislation

From the legal perspective, major countries and regions in the world have recognized
the potential problems in AI development and have introduced recommendations and
the framework for governance principles. Considering that AI development is still
in the initial period, the premature adoption of rules and regulations for mandatory
restrictions can probably erect barriers to innovation. In this regard, major coun-
tries and regions in the world are not eager for regulatory measures. By comparing
China, the US and the EU, it finds that China and the US in general take a relatively
more relaxed attitude toward AI regulations. In the draft of Guiding Principles for AI
Application in September 2019, the US paid much attention to the benefits and costs
of evaluation and regulation, as well as the flexibility of regulation. Compared with
other countries and regions, the EU obviously pays greater attention to imposing
restrictions on potential problems in AI development. In the Report on the Imple-
mentation of the Ethics Guidelines for Trustworthy AI in September 2019, the EU
stated the vision to speed up legislation and translate from “soft law” to restrictions
to be a global leader in legislation and regulations. In the 2020 White Paper, EU set
two specific goals of building an ecosystem of excellence and a unique ecosystem
of trust with specific requirements in the regulatory framework, product safety and
compliance with law enforcement. In general, the varying trends in different coun-
tries and regions are relevant to international competition, industry situations, and
values and concepts in local areas (Table 1).

3.2 Policies and Laws

3.2.1 U.S.

In February 2019, US president Trump signed Executive Order 13,859 on the new
national strategic plan of Maintaining American Leadership in Artificial Intelli-
gence, aiming to improve international advancements in AI (Fig. 1). The Order
requires Federal determined implementing agencies take active measures to imple-
ment the new national AI strategy with a focus on following objectives: (a) promoting
investment in AI R&D; (b) enhancing access to data and algorithm resources for AI
Table 1 AI Policies and Law in some countries in the world
16

Regions/countries Month/year Documents Major contents Types

Germany 10/19 Opinion of The Data Ethics Commission” of the German government released 75 concrete Policy
the Data recommendations for regulation on data and algorithmic systems. The report mainly recommendations
Ethics covers six aspects
Commission 1. General standards for data governance: fit-for-purpose data quality, risk-adequate
level of information security and interest-oriented transparency. 2. Data rights and
obligations: a contribution to the generation of data should entitle a party to specific data
rights in the form of codetermination and participation rights; these rights in turn impose
obligations on other actors.
3. Establishment and improvement of public data platforms. 4. General standards for
algorithmic systems: human-centered design, compatibility with core social values,
sustainability in the design and use of algorithmic systems, guarantee of robustness and
security, and minimizing bias and discrimination. 5. Criticality pyramid and risk-adapted
regulatory system for the use of algorithmic systems, from minor to severe: no special
measures, disclosure obligations toward supervisory bodies, ex-post controls, ex-ante
approval procedures, transparency obligations, partial or complete ban of an algorithmic
system.
6. Mandatory labeling: system operators are required to ensure transparency regarding as
to whether, when and the extent to which algorithmic systems are used
(continued)
Y. Cui
Table 1 (continued)
Regions/countries Month/year Documents Major contents Types
France 11/19 Facial Commission nationale de l’informatique et des libertés (CNIL) in this report calls for Policy
Recognition: attention to substantial risks carried by facial recognition: 1. involvement of highly recommendations
for a Debate sensitive, biometric data that are subject to special protection;
Living up to 2. a contactless and potentially ubiquitous technology; 3. an unprecedented surveillance
the potential, capable of undermining societal choices;
1 AI Policies and Laws

Challenges 4. fallible and costly technologies that require a clear and comprehensive assessment.
The report proposes although facial recognition is allowed in an experimental approach
for creating a permanent framework from the outset, the deployment of these systems
(including trials) must be subject to the modern legal framework where it is essential to
respect people and provide special protections to children. CNIL if necessary, will
impose the necessary corrective measures or even call for unlawful devices to be
removed for protections
UK 02/20 Artificial The Committee on Standards in Public Life in this report suggests the Seven Principles Policy
Intelligence of Public Life that applies to all public office-holders, including selflessness, integrity, recommendations
and Public objectivity, accountability, openness, honesty, and leadership.
Standards: Eight recommendations are made to government, national bodies and regulators to help
report create a strong and coherent governance and regulatory framework for AI in the public
sector, including: 1. ethical principles and guidance; 2. articulating a clear legal basis for
AI;
3. eradicating data bias in compliance with Equality Act 2010; 4. establishing a
regulatory assurance body to identify gaps in the regulatory landscape and provide
advice to individual regulators and government on the issues associated with AI;
5. ensuring provisions for ethical standards that providers abide by in the procurement
processes; 6.guarantee of the Crown Commercial Service’s digital market to provide AI
products and services that meet their ethical requirements; 7. Mandatory impact
assessment that should be published;
8. Government guidelines for public bodies about the declaration and disclosure of their
AI systems
(continued)
17
Table 1 (continued)
18

Regions/countries Month/year Documents Major contents Types

05/20 Explaining This cobadged guidance by the Information Commissioner’s Office (ICO) and The Alan Governance
Decisions Turing Institute aims to give organizations practical advice to help explain the processes, principles
Made with services and decisions delivered or assisted by AI, to the individuals affected by them.
AI This guidance, a recommended standard, can have a reference significance for
algorithmic legislation that is explainable in the world. Six main types of explanation for
AI decisions include explanation of rationale, responsibility, data, fairness, safety and
performance, and impact.
2. Five contextual factors have an effect on explaining AI decisions: domain you work
in; impact on the individual; data used; urgency of the decision; and audience it is being
presented to.
3. Four principles to follow, ensuring that the decisions you make using AI are
explainable: be transparent; be accountable; consider the context you are operating in (It
is about paying attention to elements that can have an effect on explaining AI-assisted
decisions, and managing the overall process.); and reflect on the impact (the value of
reflecting on the impacts of your AI system helps you explain to individuals affected by
its decisions that the use of AI will not harm or impair their wellbeing.)
(continued)
Y. Cui
Table 1 (continued)
Regions/countries Month/year Documents Major contents Types
ROK 12/19 National AI Strategy The Korean government announced the national National strategies
strategy to advance ROK from an IT power to an
AI power. It plans to upgrade its competitiveness
in the AI industry parallel to the world first rank
by 2030. Three areas of development are covered.
1 AI Policies and Laws

1. Expand infrastructure. In terms of the creation

of the AI ecosystem and R&D, the Korean
government will push ahead with comprehensive
public sector data to be made freely accessible by
2021, establish Gwangju AI park by 2024 and
invest over KRW 1 trillion (approx. USD 858
million) into the next generation smart chips by
2029.
2. Nurture start-ups. The Korean government will
foster centralized AI start-ups in the next year
with support in relaxed restrictions and improved
legal services. 3. Nurture talents. The Korean
government will set up AI majors in colleges and
universities from the next year on and courses of
software and AI will be added to education in the
high and primary schools by 2022.
In face of changes in the labor market caused by
the AI technology, the Korean government will
further expand the coverage of social security
schemes, while establishing a system for
interagency cooperation to enhance management
on the security of AI services
(continued)
19
Table 1 (continued)
20

Regions/countries Month/year Documents Major contents Types

Singapore 11/19 National AI Strategy The Singaporean strategy that lasts 11 years National strategies
mentions vision, approach, national projects and
the creation of the ecosystem for future AI
development.
5 National AI projects are identified. 1. Transport
and logistics: pooling and dynamic assignment of
trucking jobs; intelligent routing and scheduling
of trucks; AI-enabled planning and modeling for
better urban planning.
2. Smart cities and estates with more efficient and
seamless municipal services: AI-powered chatbot
for reporting municipal issues; pre-emptive estate
maintenance with sensors and AI; intelligent
insights on how residents are using facilities.
3. Healthcare on chronic disease prediction and
management: personalized risk score for chronic
diseases; clinical decision support for primary
care doctors; personalized management for
chronic diseases.
4. Personalized education through adaptive
learning and assessment: students will have a
more personalized learning experience through
adaptive learning; the automated marking system
can reduce time on routine assessment tasks for
teachers; each student will be guided by a virtual
partner as they learn.
5. Safety and security of border clearance
operations: 100% automated immigration
clearance for all travelers; AI-assisted enhanced
border security and immigration officers will
Y. Cui

achieve productivity gains

(continued)
Table 1 (continued)
Regions/countries Month/year Documents Major contents Types
Australia 11/19 Artificial Intelligence: solving problems, growing The Australian government released this report National strategies
the economy and improving our quality of life for helping develop a national AI capability to
boost the productivity of Australian industry,
create jobs and economic growth, and improve the
quality of life for current and future generations.
1 AI Policies and Laws

The roadmap depicts seven development areas in

AI: 1. talents: promoting talent training and
introducing; 2. upgrade of labor force: realizing
the vocational transformation and skill upgrading
of labor force; 3. data: enhancing data governance
and sharing;
4. society: building the trust of the society in AI;
5. scientific research: strengthening scientific and
technological research; 6. infrastructure and
security: digitalizing infrastructure and network
security; 7. regulations and ethics: setting up a
standard, sharing system and governance
principles
21
22 Y. Cui

R&D; (c) establishing principles for AI regulation; (d) training AI and the Amer-
ican Workforce; (e) strengthening international cooperation to promote American
advancements in AI technology and innovation.
Since June 2019, implementing agencies of the federal government to implement
the abovementioned order has taken active actions by introducing policies and action
plans and establishing institutes and offices to have the national strategy of AI devel-
opment and innovation proceed in all aspects at the federal level. As of mid-June
2020, progress has been made in the following six aspects.

Investments in Fundamental R&D

The first is to promote investment. In September 2019, through “the Networking and
Information and Technology Program (NITRD)” of the National Science and Tech-
nology Council, AI research investments and activities of NITRD member agencies
were planned and coordinated. Pursuant to laws, the Congress was provided with the
first NITRD Supplement to President’s Budget in FY 2020, based on priorities of the
AI R&D and other needs with a proposed investment budget of 973.5 million dollars
and a particular focus going to the AI application R&D in National Science Foun-
dation (NSF), Office of Science of Department of Energy (DOE), and U.S. National
Institutes of Health. The White House in February 2020 announced that the federal
government commits to doubling AI funding in the budget for FY 2021.
The second is by investment to drive technological breakthroughs in AI. The
National AI R&D Strategic Plan: 2019 Update published on June 21, 2019 coor-
dinates and guides all federal agencies to make technical breakthroughs by raising
investments in the next eight advantageous AI areas for the US:
● To establish the long-term mechanism for AI R&D and investment;
● To develop effective ways for human-AI collaboration;
● To understand and respond to the ethical, legal and social influences of AI;
● To ensure the safety of the AI system;
● To develop shared public datasets and scenarios for AI training and testing;
● To establish technical standards and norms to measure and assess AI technology;
● To have a better understanding of the national need of the workforce for AI R&D;
● Expanding private and public cooperation partnerships to promote AI develop-
ment.
The third is to support R&D for addressing technical barriers in building a trust-
worthy system for AI applications. To tackle a series of technical barriers identified
in the 2019 update existing in AI R&D and applications in terms of transparency,
accuracy, safety and flexibility, the federal government will prioritize investments
in R&D to promote public trust in AI technology and its applications to ensure
the development and applications of future AI technology in a way compliant with
national core values. For example, the Defense Advanced Research Projects Agency
(DARPA), through the “AI Next Campaign” initiative, invested in developing AI
 1 AI Policies and Laws

Fig. 1 Diagram of U.S. AI Policies

23
24 Y. Cui

approaches of higher data efficiency, thus reducing dependence on large-scale data

as well as the ethical and technical risks posed by frequent data usage.
The fourth is identifying agencies to coordinate cross-agency AI investment.
Considering the wide scope of federal AI R&D investment, the National Science
and Technology Council will be responsible for investment coordination among all
federal agencies. The coordination mechanism consists of three work groups:
The group of high-level leaders to make strategic planning. Its responsibility is
assumed by the “Select Committee on AI” approved by the president in May 2018.
The members include heads of agencies in charge of AI R&D with the duty of making
suggestions to the president on the AI R&D strategy.
The group of implementing agencies to pursue plans and strategies. The subcom-
mittee on Machine Learning and Artificial Intelligence (MLAI) of NSTC, comprised
of those in charge of AI in federal agencies, takes the responsibility for completing
the tasks assigned by the Select Committee on AI, implementing the national AI
strategic plan, and recognizing and tackling major policy problems in AI research,
testing, standards, education, and implementation.
The group of researchers and technical experts. The formed “Artificial Intelli-
gence Research and Development Interagency Working Group (IWG)”, comprised of
project managers and technical experts in AI R&D, is managed by the Subcommittee
on NITRD under NSTC to coordinate work plans for AI R&D across participating
agencies.

Opening of Government Resources

The federal government boasts a wealth of data, models and computing resources, a
key to AI R&D. Enhanced access to data and computing resources will involve more
experts and researchers on the AI R&D frontier and contribute to global competi-
tiveness in the industry. In other words, the US policy is to expand experts’ access
to high-quality, efficient and fully traceable federal data, models and computing
resources for AI R&D under federal government sponsorship.
The first is through expanded access to the Federal data to promote the AI R&D
and testing. On July 10, 2019, the White House released a notice to solicit public
opinions on priority access or quality improvements for Federal data and models
used in the AI R&D and testing.
According to the Federal Data Strategy 2020 Action Plan unveiled on December
23, 2019, all Federal agencies have to complete following actions in 2020:
● In January, 2020, all Federal agencies should determine the dataset with their data
prioritized to be open;
● Before June, 2020, all Federal agencies should evaluate the level of data
equipment;
● Before June, 2020, all Federal agencies should evaluate the employees’ ability of
data processing;
● In June, 2020, all Federal agencies should announce the data index catalog;
1 AI Policies and Laws 25

● Before September, 2020, all Federal agencies should confirm the priority in data
needs;
● By the end of September, 2020, all Federal agencies should establish the data
management agency.
The second is to prioritize AI R&D in the allocation of high-performance
computing resources and cloud services. According to the 2019 Update, within
the scope of laws, the federal government is allowed to allocate computing power
resources, hardware equipment and cloud services appropriately in favor of AI R&D.
For example, the largest supercomputer in the world, developed under the coopera-
tion of DOE and the National Cancer Institutes of NIH, is dedicated to seeking new
resolutions to pressing cancer challenges with the help of AI.

Removal of Innovation Barriers

While enacting and implementing relevant policies, laws and standards, the federal
government has to ensure that they support and encourage AI development and
innovation but also protect economic and national security, civil liberties, privacy,
and American values.
The first is to draw up guidelines and principles for regulations of AI applica-
tions. On January 13, 2020, the Director of the Office of Management and Budget
(OMB) released the draft of Guidance for Regulation of AI Application to request
comments, with a highlight on “encouraging innovation and growth in AI” while
“reducing barriers to the deployment and use of AI”. To ensure public participation,
prudence and flexible regulations, and advancement in trustworthy AI, ten princi-
ples are proposed for AI regulations: ➀ public trust in AI; ➁ public participation; ➂
scientific integrity and information quality; ➃ risk assessment and management; ➄
benefits and costs; ➅ flexibility; ➆ fairness and nondiscrimination; ➇ disclosure and
transparency; ➈ safety and security; and ➉ interagency and coordination.
The second is to set relevant technical standards. In response to Executive Order
13,859, the National Institute of Standards and Technology (NIST) prepared U.S.
Leadership in AI: A Plan for Federal Engagement in Developing Technical Stan-
dards and Related Tools on August 9, 2019. The plan ensures that relevant technical
standards can reflect the federal priorities set for AI innovation, public trust and
public confidence in systems that use AI technologies. Moreover, the establishment
of international standards can promote and protect these priorities.

Improvement of Government Services

The federal government has taken actions to fully tap into AI development to improve
the ability and quality of services pursuant to laws and American values under the
preconditions of secured civil liberties and privacy.
26 Y. Cui

First, efforts are made to improve government services through the collaboration
of multiple parties. Held by the White House on September 9, 2019, the Summit on AI
in Government invited leaders and experts from government, industry, and academia
to spark ideas on short-term resolutions by leveraging AI to improve government
services, such as improvement in services procedures, and counseling services to
the government AI strategy, with a special emphasis on enhancing the understanding
and trust of the private sector for the government to adopt the AI.
Second, efforts are made to promote the adoption of AI in the government
through interagency cooperation. The General Services Administration launched
the AI Center of Excellence (AI COE) in October 2019. Strategic coordination and
support from the fundamental framework of the COE can foster cooperation in AI
applications in the entire federal government, share professional knowledge and best
practices in the use of AI and improve service procedures to deliver better results.
Federal agencies are working actively to examine their own procedures, programs,
and activities to ensure that the use of AI can facilitate mission completion and
improve ways of providing services to the American people.
Third, efforts are made by federal agencies to make progress in AI policies. Since
the announcement of Executive Order 13,859, federal agencies have been very active
in implementing AI strategies. From the second half of 2019, some federal agencies
unveiled their AI and data science strategy, including the following:
● In August 2019, the Department of Defense (DoD) released the DoD Digital
Modernization Strategy, proposing through AI to accelerate changes in military
philosophy to prioritize AI development and adopt AI to empower U.S. military
information strategy transformation.
● In October 2019, the NSF announced launching National Artificial Intelligence
Research Institutes (NAIRI), planning to invest 120 million dollars in support of
large-scale AI research projects, AI research talent training and AI application
promotion. According to the released solicitation for the “NAIRI program”, this
program will be advanced under a joint government effort between the National
Science Foundation (NSF), U.S. Department of Agriculture (USDA) National
Institute of Food and Agriculture (NIFA), U.S. Department of Homeland Security
(DHS) Science & Technology Directorate (S&T), and the U.S. Department of
Transportation (DOT) Federal Highway Administration (FHWA).
● In November, 2019, the U.S. The Department of Veteran Affairs set up the National
Artificial Intelligence Institute, improving the protection level of veterans through
AI R&D and applications.
● In January 2020, the DOT and NSTC copublished Ensuring American Leadership
in Automated Technologies: Automated Vehicles 4.0. At the legal level, the report
emphasizes removing barriers that prevent the development of rules and regula-
tions for automated vehicles while proceeding with the formulation of industry
norms.
● In February 2020, the DoD adopted the Ethical Principles for Artificial Intelli-
gence, encompassing five major areas: ➀ responsible; ➁ equitable; ➂ traceable;
➃ reliable; ➄ governable. The DoD Joint Artificial Intelligence Center will be
1 AI Policies and Laws 27

the focal point for implementation coordination of AI ethical principles for the
department, guiding DoD and the U.S. military force for R&D and applications
of AI technology in a legal and ethical way.

Favorable International Environment

The U.S. values cooperation with global allies in AI applications, promotes devel-
opment and innovation in AI through international exchanges, actively participates
in the development of international rules and standards for AI, and works together
to address opportunities and challenges of common interest.
In May 2019, the US, together with 42 countries and regions, including member
countries in the Organization of Economic Cooperation and Development (OCED),
adopted the Recommendation on Artificial Intelligence. The Recommendation
suggests that AI benefits people and the planet by boosting inclusive growth, sustain-
able development and well-being; AI systems are designed in a way that respects
the rule of law, human rights, democratic values and diversity, and they include
appropriate safeguards; there is transparency and responsible disclosure around AI
systems; AI systems function in a robust, secure and safe way throughout their life
cycles with potential risks being continually assessed and managed; organizations
and individuals who develop, deploy or operate AI systems are held accountable for
their proper function in line with the above principles.
In the G20 Ministerial Meeting on Trade and Digital Economy in June 2019, the
US proposed the “human-centric AI principle” and adopted the OECD Recommen-
dation on Artificial Intelligence.
In February 2020, the US engaged in building the OECD AI Policy Observatory,
a platform to share AI policies, standards and practices in over sixty countries and
regions. It can promote exchanges in AI policies and best practices and contribute to
encouragement, fostering and supervision of responsible AI R&D and applications
among countries and regions in the world.

Exploring Governance Principles

On April 8, 2020, the Bureau of Consumer Protection of Federal Trade Commission

(FTC) published Using Artificial Intelligence and Algorithms, probably to be one
of the underpinning principles for the US to regulate AI in the future. Five specific
aspects are included. The first is to be transparent: don’t deceive consumers about
how you use automated tools; be transparent when collecting sensitive data; if you
make automated decisions based on information from a third-party vendor, you may
be required to provide the consumer with an “adverse action” notice. The second is
to explain your decisions to customers: explain the reason why you deny consumers
something of value based on algorithmic decision-making; if you use algorithms to
assign risk scores to consumers, also disclose the key factors that affect the score
and the rank is ordered for importance; if you might change the terms of a deal
28 Y. Cui

based on automated tools, make sure consumers are informed. The third is to ensure
your decisions are fair: don’t discriminate based on protected classes (races, color,
etc.); focus on inputs but also outcomes; give consumers access and an opportunity
to correct information used to make decisions about them. The fourth is to ensure
that your data and models are robust and empirically sound: make sure that your AI
models are validated and revalidated to ensure that they work as intended, and do not
illegally discriminate. The fifth is to establish the accountability mechanism: protect
the algorithm from unauthorized use and consider whether it would make sense to
use independent standards or independent expertise to step back and take stock of
AI.
At the end of May 2020, the US issued the last announcement among G7 member
countries to join the “Global Partnership on AI (GPAI)”. Virtually, it equals designing
the mechanism for cooperation and coconstruction of AI regulation principles
together with many other countries (Fig. 2).

3.2.2 EU

Following the publishing of Artificial Intelligence for Europe by EU in April 2018,

the EU Commission announced the White Paper on AI—A European Approach to
Excellence and Trust (hereafter the White Paper) in February 2020. On the basis of
AI uptake and European advantages, it proposes clear objectives and an approach
to AI development in the EU, aiming to advance AI development through constant
breakthroughs in technology, better serve European people and secure a world leading
place in AI development, particularly in supervision.
The White Paper lists three development goals for AI: improving people’s well-
being, boosting economic development, and securing public interests. Two key objec-
tives are to create an “ecosystem of excellence” by mobilizing resources to promote
AI R&D and applications and an “ecosystem of trust” with a future regulatory
framework for AI.

Establishment of an Ecosystem of Excellence

The established ecosystem of excellence can be optimized and able to support the
development and growth of AI across the EU economy and public administration.
First, work with member countries.
● The Commission will propose a revision of the coordinated plan to member states,
planning to attract over 20 billion euro investments in AI in the EU per year over
the next decade.
● The EU will make resources available from the “Digital Europe Programme”
and “Horizon Europe” as well as from the “European Structural and Investment
Funds” to support less developed regions and rural areas.
 1 AI Policies and Laws

Fig. 2 Details on the EU White Paper

29
30 Y. Cui

● By addressing societal and environmental well-being as a key principle, AI can be

leveraged to control climate change and combat other environmental degradation.
Second, constantly enhance the R&D and innovation. It is imperative to create
more synergies and networks between the multiple European research centers for
AI and strengthen cooperation in talent introduction, technology development and
other aspects. Equally important is to create world reference testing centers.
Third, highlight skill education and training. Support will be given to estab-
lish the Digital Europe Programme networks comprised of leading universities and
higher education institutes. It hopes to attract top researchers while offering world-
leading master’s programs in AI. It is necessary to upskill sectoral regulators and
the workforce to let everyone become fit for AI-led transformation and benefit from
digitalization.
Fourth, support the establishment of innovation hubs.
● The Commission will work with member states to ensure at least one digital
innovation hub per member state.
● The Commission and the European Investment Fund will colaunch a pilot scheme
of 100 million Euros and scale it up significantly to provide equity financing for
innovative developments in AI.
Fifth, promote the public–private cooperation. In the context of “Horizon Europe”,
the Commission will set up a new public–private partnership in technologies of AI,
data and robotics.
Sixth, the adoption of AI in the public sector should be promoted. The Commis-
sion will initiate dialogs with different sectors, in particular those with healthcare,
rural administrations and public service operators, to present an action plan for facil-
itating AI development, experimentation and adoption. The “Adopt AI Programme”,
which will support public procurement of AI systems, can help to transform public
procurement processes.
Seventh, improve data computing power. The Commission has proposed more
than 4 billion euros under the “Digital Europe Programme” to support high-
performance and quantum computing, including edge computing and AI, data and
cloud infrastructure.
Eighth, international exchanges and cooperation should be strengthened.
● When developing its ethical guidelines, the High-Level Expert Group has involved
a number of non-EU organizations and several governmental observers.
● The EU was closely engaged in developing the OECD’s ethical principles for AI.
● The EU has participated in other multilateral forums on AI, such as participation
in the follow-up of the report of the High-Level Panel on Digital Cooperation.
● The Commission will continue to cooperate with like-minded countries, in partic-
ular tightly monitoring the policies of third countries that limit data flows, but will
also address undue restrictions in bilateral trade negotiations and through action
in the context of the World Trade Organization.
1 AI Policies and Laws 31

Establishment of an Ecosystem of Trust

According to the White Paper, risks posed by AI are mainly in three aspects:
(1) Risks in personal data, privacy protection, nondiscrimination and other funda-
mental rights.
(2) Risks in safety, such as autonomous vehicles erring in recognizing roadblocks.
(3) Risks in the liability regimes. Current EU legislation becomes unclear when
allocating responsibility for AI product safety, causing uncertainty in AI product
applications and law enforcement.
These risks will prevent people from building trust in AI. Such loss of trust, in
addition to a lack of investment and skills, is the main factor holding back the broader
application of AI. It is necessary to establish an ecosystem of trust for improving a
regulatory and legal system that can effectively prevent risks and secure fundamental
rights.
First, protect rights and identify liabilities. The use of AI should protect funda-
mental rights, including the rights to freedom of expression, freedom of assembly,
human dignity, nondiscrimination and protection of private life.
Second, amend laws on product safety and liability. It is an open question whether
stand-alone software is covered by EU product safety legislation. General EU safety
legislation currently in force does not apply to services and therefore, in principle,
not to services based on AI technology. The required specific provisions include
conducting a new risk assessment for the AI system; considering the liability to
the producer for mental safety risks of users; providing specific requirements for
addressing safety risks in the use of faulty data at the design stage; and making
algorithms transparent. In the Report on the safety and liability implications of arti-
ficial intelligence, the Internet of Things and robotics in February 2020, the EU
Commission recommends clear amendments to the product liability directive:
● To further identify the concept of producers with explicit obligations and clarify
the scope of obligations and responsibilities for the damages caused by defects in
software or other digital functions;
● To mitigate the burden of proof on victims and establish targeted common liability
rules at the national level;
● To modify the current “put-into-circulation” concept, it is necessary to clarify
who will ultimately be liable for changes to the product if the changes are caused
by autonomous behaviors of AI systems. In addition, it proposes identifying
network safety obligations, making operators clear about their obligations to avoid
assuming unnecessary liabilities.
Third, identify the scope of the regulatory framework.
● To identify the concept of “artificial intelligence” for a reasonable regulatory
scope.
● To follow a risk-based approach, based on risk levels to differentiate between
proportionate regulatory interventions, from the industry and expected usage,
32 Y. Cui

safety protection, consumer rights and fundamental rights to evaluate risk levels,
with a focal point on solving the high-risk problem.
● The use of AI applications for recruitment processes as well as in situations
impacting workers’ rights for the purposes of remote biometric identification and
other intrusive surveillance technologies would always be considered “high-risk”.
Fourth, clarify the legal obligations of all parties. The regulatory measures for AI
include: training data should be extensive to avoid discrimination and protect privacy;
it should keep a record of algorithms and training data to enhance transparency and
traceability; it should volunteer to offer information on the use of the high-risk AI
system, including the explanation of its limitations and risks, informing the public
that they interact with is the AI system not human beings; it should develop AI in a
responsible way to effectively prevent and control risks; it should not depend solely
on AI and human supervision should be enhanced; there should be restrictions on
remote biometric identification.
Fifth, reasonably determine the subject of liability. The AI system involves
multiple parties, including developers, producers, distributors, importers, service
providers and users. Therefore, each liability should be assumed by those who are
best able to deal with the potential risks. Developers will be the most suitable to
handle risks at the development stage, but their risk prevention and control abilities
at the use phase may be more limited. The EU requirements for liable parties apply
to all economic operators providing AI-enabled products and services in the EU,
irrespective of whether they are established in the EU.
Sixth, strengthen compliance and law enforcement. Applicable legal requirements
must be effectively enforced by both competent national and European authorities.
Competent authorities should be in a position to investigate individual cases and
to assess the impact on society. The Commission can carry out an objective, prior
conformity assessment of high-risk applications, including examining algorithms
and datasets used in the development phase.
Seventh, encourage nonmandatory compliance voluntarily. For AI applications
that do not qualify as “high-risk” and that are therefore not subject to mandatory
requirements, in addition to applicable legislation, an option would be to establish a
voluntary labeling scheme, under which nonmandatory-covered actors on a voluntary
basis could decide to make themselves subject either to legal requirements or to a
specific set of similar requirements especially established for the purposes of the
voluntary scheme.
Eighth, improve governance on AI comprehensively.
● A European governance structure on AI in the form of a framework for cooperation
of national competent authorities could have a variety of tasks as a forum for a
regular exchange of information and best practice, identifying emerging trends,
advising on standardization activity and on certification. It should also play a key
role in facilitating the implementation of the legal framework, such as through
issuing guidance, opinions and expertise.
● The governance structure should guarantee the maximum participation of
stakeholders.
1 AI Policies and Laws 33

● It should establish close links with other EU and national competent authorities in
the various sectors. The carrying out of conformity assessments could be entrusted
to notified bodies designated by member states.

Focus on Facial Recognition

The White Paper has given adequate consideration to biometric identification tech-
nology. In facial recognition technology: Fundamental Rights Considerations in the
Context of Law Enforcement in November 2019, the European Union Agency for
Fundamental Rights points out the threats imposed by facial recognition technology
(FRT). The following recommendations are proposed to improve legal restrictions:
● To strengthen legal requirements. The clear and detailed legal framework should
be sufficient to regulate the use of FRT.
● To differentiate use purposes. A distinction must be made between the processing
of facial images for verification purposes, when two facial images are compared to
verify if they appertain to the same person, and their processing for identification
purposes, when a facial image is run against a database or watchlist of facial
images. The risk of interferences with fundamental rights is higher in the second
case; therefore, the necessity and proportionality test must be stricter.
● To emphasize the control of FRT usage in public places. When facial images
are extracted from video cameras deployed in public spaces, the necessity and
proportionality of performing facial recognition should be evaluated.
● To reduce error risks. When deploying the technology, the risks of wrongly
flagging people must be kept to a minimum.
● To conduct constitutional review. To carry out a fundamental rights impact
assessment, public authorities have to obtain all necessary information from the
industry, which is required to assess the technology’s impact on fundamental
rights. In February 2020, Margarethe Vestager, Executive Vice President of the
EU Commission, told the reporter in an interview that as remote FRT breached
the terms of approval or authorization provided in the General Data Protection
Regulation, EU is considering regulating facial recognition by legislation (Fig. 3).

3.2.3 Russia

On October 10, 2019, President Putin signed Executive Order on the Development
of AI in Russia to “approve the National Strategy for the Development of Artifi-
cial Intelligence for the period until 2030 (hereafter Strategy), aiming to accelerate
AI development, conduct scientific research in the field, increase the accessibility
of information and computing resources for users, and improve personnel training
system”. The strategy lists one overarching goal, seven principles, six major tasks and
three implementing mechanisms for Russian AI development in the coming decade.
It is designed to boost advancement in AI development and applications but also
 34

Fig. 3 Analyses of the AI Strategy

Y. Cui
1 AI Policies and Laws 35

seeks a world leading position in AI for ensuring national security and improving
economic power and people’s well-being.
The order requires that the Russian government incorporate AI development
strategy documents into the national program of “National Digital Economic Devel-
opment Programs in Russia” before December 15, 2019 and submit a report on the
implementation of the strategy to the president every year.

One Overarching Goal

The strategy proposes the overarching goal for AI development: to improve people’s
well-being and living standards, safeguard national security and the rule of law, and
enhance economic sustainability and competitiveness to make Russia a world leader
in AI.

Seven Basic Principles

While developing AI and advancing the implementation of the strategy, seven

principles should be followed:
● To protect human rights and freedoms. The rights and freedoms of men and citi-
zens granted in Russian laws and international law should be protected, including
the right to work and the opportunity that individuals can acquire knowledge and
skills to become fit in the digital economy.
● To secure safety and reliability. The use of AI is prohibited to hurt natural and
legal persons in the prevention and avoidance of negative effects that AI can bring.
● To guarantee transparency and fairness. Make sure that the working program,
algorithms and results of AI are known and predictable, and algorithms will not
discriminate users.
● To emphasize technology sovereignty. It should ensure that Russia can inde-
pendently develop AI by using domestic technology and solutions in as many
scenarios as possible.
● To encourage innovation and collaboration. By putting AI at a key place in
economic development, it should ensure close cooperation among all social
sectors in AI.
● To conserve resources reasonably. The currently in-force technology policies of
the government should be prioritized in implementation and adjustment to boost
AI development.
● To support market competition. The market relationship in AI development should
be invigorated, and behaviors that dampen competition among market entities
should be prohibited.
36 Y. Cui

Six Major Tasks

Task one, to support fundamental and applied scientific research on AI. Priority is
given to the following three fields: algorithms that simulate the biological decision-
making system; autonomous learning and the active adaptability of algorithms to the
new targets; and autonomous decomposition of complicated tasks and their active
resolutions. Key areas include:
● To keep long-term and favorable support to fundamental scientific research with
increased technology R&D staff;
● To arouse the interests of individuals and enterprises in AI R&D;
● To support cross-disciplinary research on AI in all economic sectors;
● To enhance patent research on AI at home and abroad with a regular update of
the latest technology achievements;
● To enlarge the infrastructure scale of the fundamental research by providing more
computing resources, databases and datasets to scientific researchers;
● To develop international cooperation through international conferences and other
channels to promote technical exchanges among domestic experts and overseas
counterparts;
● To improve the efficiency in the evaluation of R&D breakthroughs and their
applications.
The targets for this task are to achieve a significant increase in the number of
articles on research achievements by Russian scientists in top international journals,
AI patents and applications of AI-assisted solutions by 2024.
Task two: to develop and promote AI-assisted software. Key areas include:
● Creating comfortable working conditions for AI experts by providing necessary
fund support and encouraging them to participate in international exchanges on
technology.
● To improve conditions for AI experts conducive to engagement in building the
open source AI library at home and abroad;
● To make universal standards for software quality, safety, compatibility and
evaluation of efficiency.
Targets are including the following: by 2024, substantial progress is made in
the extent AI experts initiate international technology exchanges and international
broadly applied software and open source data libraries are built; before 2030, devel-
oped software has to offer effective solutions in wide fields, able to take a leading
position in the international market.
Task three aims to improve the accessibility and quality of data required in AI
development. Key areas include:
● To develop a unified and constantly updating approach to describe, collect and
label data and establish information and communication infrastructure able to
support access to data.
1 AI Policies and Laws 37

● To create and upgrade all kinds of public data access platforms and secure access
priority to government institutes and social organizations;
● To establish an effective legal regulatory framework so that the collection, flow
and accessibility of data is ensured pursuant to municipal law and the data obtained
from scientific research and economic activities are stored in the territory of
Russia.
By 2024, datasets in compliance with collection and identification standards, as
well as public open data platforms, should provide access services to organizations in
Russia; personal and other data should be well protected in accordance with law; and
data processing must honor the international commitment by Russia. Before 2030,
all types of datasets opened in public platforms should be sufficient to support the
R&D of crucial AI programs.
Task four: to improve the utility of hardware in AI development. Key areas include:
● To develop fundamental research on the computing system architecture;
● To provide government support to the development and ownership of intellectual
property rights to the domestic high-speed energy-saving processor as well as the
software and hardware using domestic optoelectronic components;
● To provide government support to the establishment of the software and hardware
composite system comprised of domestic optoelectronic components;
● To support advanced centers of electronic component R&D, assembly and test
production;
● To establish and develop high-performance data processing centers with favorable
access for scientific researchers and experts.
By 2024, it should build infrastructure supporting domestic organizations to carry
out AI R&D, including high-performance data processing centers, while ensuring
that the performance of domestic microprocessors in terms of their speed and
energy conservation is no worse than that of similar products in other countries.
By 2030, domestic microprocessors should be widely applied in the international
market; special data processing centers based on domestic microprocessors should
be launched; intelligent infrastructure has been commercialized; R&D of the new-
type computing system has been completed; and relevant intellectual property rights
have been registered.
Task five aims to improve the supply of AI talent and raise public awareness of
AI. Key areas include:
● To introduce programming, data analysis, machine learning and other modules in
education and training programs at all levels;
● To attract enterprises and institutes in the AI cause to participate in AI promotion
and professional education activities;
● To improve interdisciplinary education quality in maths, natural science and social
science, and humanities and encourage interdisciplinary students to carry out
in-depth research on these disciplines;
● Increase the number of competition activities that foster intelligence and creativity,
such as Soviet Student Olympiad for students;
38 Y. Cui

● To create an incentive mechanism for supporting personnel to learn AI profes-

sional knowledge and skills;
● To attract top foreign scientists to come to Russia for AI R&D;
● To promote AI achievements and safety to public and social organizations and
provide relevant training and retraining programs.
By 2024, Russia should substantially increase the number of AI talents with
professional knowledge and create conditions in salary and working environments
conducive to the work of AI experts. By 2030, training programs at the world level
designed for top experts and management staff should be launched; the higher educa-
tion institute in Russia must take a world leading position in AI education; AI talent
shortages can be addressed by introducing foreign experts and other ways.
Task six establishes a comprehensive management system to coordinate the
relationship between AI and all social sectors. Key areas include:
● To establish legal systems in favor of access to public and personal data that were
initially anonymized for AI R&D and applications by setting up legal pilot zones
and other ways;
● To create legal systems and programs corresponding to simplifying the testing
and introducing procedure of AI solutions;
● To remove the administrative barriers in the export of AI civil products and
services;
● To unify standards for the use and evaluation of AI solutions and develop interna-
tional cooperation on standardization and identification of relevant products and
services;
● To improve the mechanism for public–private cooperation and investment and
mobilize private capital to AI R&D and applications;
● To set ethical norms for human-AI interactions.
Before 2024, the foregoing legal system should be established to promote the
implementation of AI strategies and address relevant problems. Before 2030, the
role of a flexible legal regulation system must be given play to safeguard public
security and advance AI technology development.

Three Implementing Mechanisms

First, enhancing organization and coordination. AI development requires the partic-

ipation of the entire society in Russia. “AI national programs” and relevant action
plans should be incorporated into National Digital Economic Development Programs
in Russia. The Digital Economy of the Government Commission on the Use of IT
to Improve the Quality of Life and Business Condition is responsible for overall
guidance, coordination and supervision of the implementation of the Strategy.
Outside the government, an association (APKIT) has been formed with members
in the AI industry from the scientific and technological circle as well as the indus-
trial and business circle. Their responsibility is to coordinate and organize scientific
1 AI Policies and Laws 39

and technological institutes and enterprises to participate in AI R&D and appli-

cations, help the government make relevant policies and action plans, supervise
implementation and participate in result evaluation.
Second, providing fund guarantee. The money for implementing the strategy will
be financially supported by the Russian Federation budget, the extrabudgetary funds
of each district and city (including funds from state-owned enterprises) and private
investments.
Third, report and adjust mechanisms on a regular basis. The government should
report the implementation of the strategy each year to the president. The strategy will
be adjusted every three years based on adjustment suggestions by the government
and union after the president signs the executive order.

3.3 Comparison of AI Governance Principles of China

and Other Countries

In June 2019, China introduced the governance principles for developing respon-
sible AI, with eight recommendation guidelines in harmony and human friendli-
ness, fairness and justice, inclusiveness and sharing, respect for privacy, safety and
controllability, shared responsibility, open and collaboration and agile governance.
In January 2020, the US requested comments on the draft of Guidance for Regu-
lation of AI Application, proposing ten principles: public trust in AI; public partic-
ipation; scientific integrity and information quality; risk assessment and manage-
ment; benefits and costs; flexibility; fairness and nondiscrimination; disclosure and
transparency; safety and security; interagency and coordination.
In April 2019, the EU took a step ahead to release Ethics guidelines for trustworthy
AI, putting forward principles and directions for AI governance from human agency
and oversight, safety, privacy and data governance, transparency, inclusiveness,
societal well-being and accountability (Table 2).
AI governance principles in China, the US and the EU can be decomposed
into ten elements: societal well-being, fairness, collaboration, good governance,
transparency, safety, privacy, accountability, data quality and human autonomy.
Of the ten, China, the US and the EU take the same side that social well-being,
fairness and human autonomy are the key principles in AI governance. In terms
of collaboration, China places greater emphasis on both domestic and international
sharing, the US values internal coordination among domestic agencies, and the EU
values international collaboration. With regard to good governance, China focuses on
agility, the US focuses on flexibility and profit and loss balance, and the EU focuses
on the ability of human oversight. In response to transparency, China makes it an
item under safety rather than an independent section, the US highlights appropriate
transparency and disclosure, and the EU hopes high-level transparency and explain-
ability. In privacy, China attaches importance to protecting underlying data, the US
sets it as an item under safety, and the EU emphasizes restrictions on data collection
40 Y. Cui

Table 2 Comparison of AI Governance Principles among China, the US and the EU

Principles China US EU
Social well-being Harmony and human Public trust Social well-being,
friendly, inclusiveness diversity
and sharing
Fairness Fairness and justice Fairness and Fairness as the core,
nondiscrimination scattering in all
principles
Collaboration Open and collaboration Interagency International
(domestic and coordination (more collaboration and
international) domestic) consensus (more
international)
Good Governance Agile governance Flexibility, break even Guarantee of human
point, risk tradeoff oversight
Transparency Transparency as a Appropriate disclosure High-level
requirement for safety and transparency transparency,
traceability and
expalinability
Safety Transparency, Focus on safety, Stability, accuracy,
controllability, stability evaluation of risks, replicability
appropriate response
Privacy Attention to protections Privacy as a Protections of data
of data acquisition requirement for safety acquisition and
usage
Accountability Coresponsibility for (Not mentioned Audit, negative
researchers, developers independently) reports and others
and users specific measures
Data quality (No mentioned) Data quality and Data quality and
scientific integrity integrity
Human Autonomy Prerequisite to respect Autonomy and public Human autonomy
people trust and human
oversight

and use. From accountability, China puts a focus on co-shouldering responsibility,

the US does not mention it independently, and the EU proposes more detailed audits,
negative reports and other specific measures. In terms of data quality, China does not
mention it, while both the US and EU have considered the quality.
In conclusion, AI governance in China, the US and the EU is consistent in overall
governance direction but different in specific governance orientations. With more
emphasis on regulating the bottom line, the US hopes to promote AI development
by resolving problems with AI regulation. The EU remains vigilant to the potential
problems AI can bring to society, hoping to reduce risks by efficient oversight.
Generally, governance in China is somewhere in between those in the US and EU.
Chapter 2
Theoretical Research on AI and the Rule
of Law

Yadong Cui

1 Overview

Theoretical research on the rule of law of AI aims to integrate the forces of law, ethics,
sociology and other related disciplines to promote the construction of the legal system
and institutional system of AI and to guarantee the safe, reliable, controllable and
healthy development of AI. This part analyzes theoretical research on the rule of law
of AI from July 2019 to June 2020 in six areas.
First, basic research on the rule of law theory of AI is analyzed. With the vigorous
development and application of AI, there have been an increasing number of legal
issues involving AI, and the discussion and research on these issues have been contin-
uously deepened and expanded. For example, the issue of automation administration,
criminal liability of AI, big data evidence, data utilization and privacy protection,
and so on. This part focuses on the core areas of theoretical research on the rule of
law of AI and selects representative views of experts and scholars in jurisprudence
and law to show the basic overview of the research in 2019.
Second, a special analysis of the research monographs on the rule of law of AI is
conducted. With “AI” and “rule of law” as key words, this part obtained all mono-
graphs on the rule of law of AI published in China in 2019 by searching Dangdang,
Taobao and Amazon. The data were integrated according to the author, author’s
nationality and publishing house to form a complete database. On this basis, the
contents, authors and publishers of the monographs are analyzed to show the general
situation of the analysis of the monographs on the rule of law of AI in China in 2019.
Third, a special analysis is made on the scientific research projects of the rule of law
of AI. Scientific research projects are a series of unique, complex and interrelated
activities to carry out scientific and technological research, which are completed
according to certain specifications within specific times, budgets and resources.

Y. Cui (B)
Shanghai Law Society, Shanghai, China
e-mail: [email protected]

© Shanghai People’s Publishing House 2022 41

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2_2
42 Y. Cui

Compared with the research on the overall change of the development of the rule
of law of AI in the last ten years in the Blue Book on AI (2019), this part focuses
on 2019 and takes the data of national social science fund projects and provincial
and ministerial-level law projects (Ministry of Justice, China law society, Ministry
of Education) as samples to analyze and compare with 2018.
Fourth, a special analysis is made of the forums on the rule of law of AI. A
forum (meeting) is an organized deliberative activity on a defined subject matter at
a defined time and place and in accordance with certain procedures. This section
collects forums on the rule of law of AI in China from July 2019 to June 2020. By
systematically sorting out the dimensions of the time, region, organizer and theme
of these forums, the hotspot issues and latest trends of the development of the rule
of law of AI are analyzed.
Fifth, a special analysis of research papers on the rule of law of AI is conducted.
This part selected all the journal articles published in 2019 and finally formed a
comprehensive and scientific database through data correction and data verification.
Then, the data were integrated according to authors, institutions, regions, cited arti-
cles and downloads. By comparing it with the data in 2018 and previous years, a
three-dimensional analysis was carried out with core authors, core journals of law,
discipline distribution and high-frequency hotspot issues as focuses.
Sixth, a search and analysis of the rule of law of AI research in major countries
around the world is conducted. This part takes Web of Science as the data acquisition
channel, searches with “Robot” and “Law”, selects the data in 2019, and focuses on
analyzing the research direction, author distribution, country/region characteristics
and other dimensions of the rule of law of AI in China, the United States, France,
South Korea, Canada, Iran, Britain, India, Australia and other countries. The research
from July 2019 to June 2020 is compared with previous years to predict the future
development trend of the rule of law of AI research in the world.

2 Theoretical Research on the AI and the Rule of Law

in China

2.1 Analysis of the Focus of AI and Rule of Law Research

2.1.1 The Legal Subject Status of AI Applications

Regarding whether artificial intelligence possesses the status of legal subjects, some
scholars believe that artificial intelligence has human-like intelligence but has not
developed human rationality, and therefore, it cannot legislate for itself or obtain legal
subject status similar to natural persons. Although artificial intelligence harbors some
powerful abilities that surpass humans, it is of no real benefits to establish a legal
subject for it. In other words, it cannot be endowed with the status of legal subjects,
similar to a legal person. The superintelligence of artificial intelligence contains
2 Theoretical Research on AI and the Rule of Law 43

tremendous risks, which must be put under humans’ domination and control; thus,
artificial intelligence can only be the object, rather than the subject, of legal relations.
In view of the intelligence and autonomy of artificial intelligence, highly intelligent
artificial intelligence can be treated as a special object and subject to special legal
regulation. Some scholars hold that attributing legal personality to intelligent robots
is reasonable from the perspective of jurisprudence, since they also have intelli-
gence, which is essential to legal persons. Attributing legal personality to intelligent
robots plays a positive role in adapting to social development, although it may have a
significant impact on traditional jurisprudence. With a view toward the future devel-
opment of criminal law, recognizing that “an intelligent robot is also a person” in
legal concepts or systems can make it easier to deal with various “crimes” and sophis-
ticated social problems involving intelligent robots more reasonably and effectively.
Other scholars hold the view that “rationality” should be regarded as the fundamental
basis for intelligent robots to enjoy the status of legal subjects. The classification of
the legal status of intelligent robots as civil law subjects can refer to the classification
of the legal status of natural persons in China. Taking the degree of rationality as
the basic classification standard and specifying “rationality” as the degree of intel-
ligence and rationality of intelligent robots, they can be classified into three basic
types: robots with civil incapacity, robots with restricted civil capacity and robots
with full civil capacity. According to the different legal statuses of intelligent robots,
the civil rights, obligations and responsibilities of robots are defined. Other scholars
believe that to address the governance dilemma brought about by the advancement
of artificial intelligence algorithms, a sound regulation system is needed. However,
the underlying reason lies in the identity crisis of human beings. With the develop-
ment of science and technology, the particularity of humans at the objective level has
gradually weakened, and the simulation of human thinking by artificial intelligence
technology is constantly approaching a true person. The philosophical approach of
monism may lead to the failure of laws to regulate artificial intelligence. Free will
is the basis of clarifying human subjectivity in legal norms. Rebuilding the standard
of goodness with the human-oriented goal is the foundation for laws to guarantee
human autonomy and carry out artificial intelligence governance.

2.1.2 The Automated Administration of AI Applications

The application of big data and AI in administrative law enforcement refers to admin-
istrative activities with the assistance of machines, data and the Internet, where admin-
istrative decisions and procedures can be visualized, realizing automatic administra-
tion. Some scholars believe that automated administration mainly includes three
types, namely, the collection and processing of information and data, the electron-
ization of administrative activities and making administrative decisions directly. It is
different from traditional administrative activities in subjects, procedures and discre-
tion, resulting in potential legal risks in the future, such as the risk of error in infor-
mation, data collection and processing, the risk of intrusion on personal privacy
and data protection, and the risk arising from the blackbox algorithm. In addition,
44 Y. Cui

there are different levels of automated administration. Similar to traditional traffic

capture, new administrative approaches such as “unattended automatic examina-
tion and approval” and “examination and approval within a second” also belong to
automated administration but obviously at a different level. Using certain criteria,
automated administration can be classified into several different levels. The realiza-
tion of a new level brings about new legal changes. Partly automated administration
represents the emergence of it from scratch and brings legal challenges in deter-
mining the nature of automatic administration. The fully automated administration
without capacity for discretion realizes administrative activities hands-free, and legal
issues also arise to be solved in the aspects of discretion determination, supervision,
responsibility and applicable situations. The next revolution of automated decision-
making with the capacity of discretion will be ushered in. We need to be prospected
and prepared.
In automated administration, the integrity and accuracy of data collection directly
determine the quality of administrative decisions. The development and application
of big data technology has effectively promoted the interconnection between infor-
mation and administrative decisions. However, while big data provides a large amount
of information for administrative decision-making, it is easy for information to be
artificially used and edited to mislead administrative decision-making. Therefore,
legislation should be regulated according to the extraction, application procedures
and legal consequences of big data. Only by regulating the use of big data by law
can the accuracy and efficiency of administrative decision-making be truly realized.
The application of big data and AI, on the one hand, gives more convenience to
the government in information publication, making it more transparent, on the other
hand, places higher requirements for the rule of law of information publication. Some
scholars believe that to resolve the phenomenon of “reluctance, fear and injustice” in
government data disclosure by means of the rule of law, we should uphold the concept
of inclusive development and develop a legal system that abides by principles such as
combining material and spiritual incentives, safe harbors, imputation of liability for
violations, free access and equal openness. To solve the phenomenon of “inability,
low speed, low quality”, we should follow the concept of cooperative governance
and build a legal system that abides by the principles of free participation, “three
divisions”, data sharing, integrity and consistency.

2.1.3 Criminal Responsibility in AI Applications

For the question about whether the intelligent robot should be treated as the subject
of criminal responsibility, there are two divided opinions in criminal jurisprudence:
denial or confirmation. Opponents hold that no intelligent robot should be granted to
such right and responsibility. As intelligent robots are a tool subordinate to humans,
even in the situation when AI autonomy causes a severe detrimental effect in society,
it is still the relevant natural persons or organizations that should be held account-
able rather than regarding AI as a criminal. Instead, proponents argue that as highly
humanoid intelligent robots have the reason and the capacity for performing acts
2 Theoretical Research on AI and the Rule of Law 45

autonomously, that is, with the ability of recognition and control required to be
the subject of criminal responsibility, intelligent robots should be recognized as the
subject of criminal responsibility. Some scholars approve the significance of the
current discussion on whether strong intelligent robots are the subject of criminal
responsibility. On the other hand, they have identified defects in the denial arguments:
that the strong intelligent robots are different from the current subject of the criminal
responsibility should not be the reason to negate the legal entity of intelligent robots;
the statement of strong intelligent robots without free will has not been confirmed;
for the idea that punishment cannot be inflicted leading to the denial of strong intel-
ligent robots as a legal entity, the causation is reversed; the recognition of the legal
entity for strong intelligent robots will not excuse the current legal subjects from
the criminal responsibility. If strong intelligent robots with recognition and control
abilities are regarded as the subject of criminal responsibility, it is reasonable on its
own but conducive to fulfilling the function of criminal law.1 With respect to account-
ability, conduct with severe adverse effects on society by intelligent robots within
the scope of their design and written program actually extends from the behaviors of
developers and users in the era of weak AI. Hence, developers and users should bear
criminal responsibility. Different from traditional crimes, the proportion of criminal
responsibility for one developer to another, for one user to another, or for developers
to users will change with intelligent robots being smarter. The growth of intelligence
may probably impact the severity of sentencing, even the judgment on the nature of
acts by doers. When intelligent robots are under the control of their own autonomy
and will conduct acts harmful to society that exceed the scope of their design and
written program, intelligent robots should be convicted and punished as a subject of
criminal responsibility. In this situation, intelligent robots together with users may
constitute joint offenses but not with developers.2 In foreign countries, regarding
whether AI can constitute a crime, theories in legal research include “psycholog-
ical elements”, “intelligent agent”, “comparison by legal human”, “legal entity” and
“the main body”. Some believe the philosophical foundations for AI as the criminal
subject are scientific positivism and moral dualism, and its realistic condition is to
have legal personality. Intelligent agents can become moral agents. Ways to develop
AI morality include approaches of “realism”, “relationship theory” and “epistemol-
ogy”, among which the “realism approach” is relatively reasonable. AI subject to
punishment can not only pave the way for holding other subjects liable but also
achieve the purpose of depriving them of their criminal capacity and generating
additional value.3

1 See Liu Xianquan. 2019. Response to Denial of Subject Status of Criminal Responsibility of
Smarter Intelligent Robots. Law Review (05).
2 See Liu Xianquan. 2019. The Path of Criminal Regulation of Artificial Intelligence Crime. Modern

Law Science (01).

3 See Peng Wenhua. 2019. Regulation of Criminal Law for Artificial Intelligence. Modern Law

Science (05).
46 Y. Cui

2.1.4 Competence of Evidence with AI Applications

AI development facilitates the use of big data as evidence. The legal definition of big
data evidence in countries has different theoretical and practical arguments, mainly
classifying such evidence as the appraisal opinion, expert opinion, or witness testi-
mony. Given the professional and scientific characteristics, big data evidence in
compliance with reality in China should be grouped into the appraisal opinion. To
review and judge such evidence, China should seek innovation with a focus on the
rules of authenticity and relevance. Rules include a “big” authenticity rule for the
purpose of the truthfulness of the big data itself; an authenticity rule to decide the
credibility of the machine algorithm for the purpose of the truthfulness of the big data
analysis results; and a double-relevance rule for the data analysis from the holistic and
specific perspectives, in particular the relevance rule that transcends human empirical
judgment, for the purpose of the relevance between the big data analysis results and
following conclusions.4
When AI is applied to the evidence judgment in criminal justice, the auxil-
iary, boundedness and rebuttable principles should be followed. In the judgment of
evidence competence, AI cannot make a material judgment, but it can check whether
the evidence is formally legal. In probative force judgment, AI cannot assume the duty
on its own, but it can play an effective auxiliary and reference role, such as finding
obvious contradictions or form changes of evidence. In judging the standard of proof,
AI cannot bear the responsibility of fact finding and proof standard judgment, but
the judgment of AI on evidence specification can guarantee judicial personnel to
obtain and review evidence as fully as possible. In the investigation, prosecution and
trial, the target, mode and focus of AI applied in evidence judgment during each
stage were different. Therefore, the AI system design should be discriminated for
the purposes of each stage.5 Due to structural deviation in obtaining evidence, one
party is unable to obtain electronic data held by the other, thus impeding the party
without data to submit claims and proofs in litigation. Therefore, it is necessary to
conduct an in-depth analysis of the attributes of electronic data and the applicable
evidence method. It can also learn from the pattern that other countries or regions
adopt to solve electronic data deviation to improve our rules related to the obligation
to submit evidence in electronic data.6

4 See Liu Pinxin. 2019. On Big Data Evidence. Global Law Review (01).
5 See Zong Bo. 2019. The Analysis of The Use of Artificial Intelligence in Criminal Evidence
Judgment. Science of Law (Journal of Northwest University of Political Science and Law) (01).
6 See Gao Bo. 2019. Solving the Problem of Asymmetry of Electronic Data: Reflection on the Basis

of Rules of Obligation to Submit Evidence. Science of Law (Journal of Northwest University of

Political Science and Law) (02).
2 Theoretical Research on AI and the Rule of Law 47

2.1.5 Privacy Protection in AI Applications

Data are the basis of AI, and algorithms are the essence of AI. The more intelligent
AI is, the more dependent it depends on the support of data and algorithms, and the
more likely it is to trigger a severe privacy crisis. In this situation, the traditional
legal framework for privacy protection is not adequate enough to provide effective
protection; neither can it maximize the value of personal information to the most.
In the face of new challenges posed by AI, privacy protection requires efforts to
improve the legal system, value the importance of technology, explore the market
mechanism and establish ethical principles.7
Some advocates introducing legitimate interest exemption to settle the conflict
between the legitimate use of personal information and privacy protection. In a
broad definition, legitimate interests refer to the legal use of interests in all manners.
Nevertheless, before applying this principle, a balance test must be conducted by
data controllers. If the test can prove that the legitimate interests of data use override
the personal interests of data owners, then the legitimate interests exemption can be
applied. The balance test can be performed in a case-by-case manner while following
the principles of necessity, purpose limitations and proportionality. In addition, data
controllers should document the whole process of the balance test, having it subject
to the supervision of data owners, government data protection authorities and courts.8
In settling limitations of private law and establishing public order for data protection,
legal research is highly dependent on traditional legal theories (in particular private
law) in a rigid way, while rather lacking innovation in theories and the legislation
mode for “data”, a new concept in law. As a pure public good, data inherently follow
the principles of mutual benefits and sharing. On this basis, the thinking pattern for
the legal theory of data should be changed from scarce-based law, the protection
of private interests, the strengthening of data control to abundance-based law, the
protection of public interests, and the modesty principle of data control. The systemic
data public order in public law should be established by narrowing the grounds for
data control.9 In data use and privacy protection, scholars in China attempt to strike a
balance between the protection of personal interests and the value of public interests
to maximize common interests.

2.1.6 Intellectual Property Protection in AI Applications

The enhancement of AI’s ability to generate intellectual achievements has brought

challenges to the traditional theory of intellectual property. Regarding whether AI

7 See Zheng Zhifeng. 2019. Privacy Protection in the Age of Artificial Intelligence. Science of Law
(Journal of Northwest University of Political Science and Law) (02).
8 See Xie Lin. 2019. Legitimate Interests Exemption for Personal Data Processing in the Big Data

Era. Tribune of Political Science and Law (01).

9 See Mei Xiaying. 2019. Between Sharing and Control The Limitations of Private Law and the

Building of Public Order in Data Protection. Peking University Law Journal (04).
48 Y. Cui

can be the subject of intellectual property rights, some believe AI has satisfied some
conditions to be the “subject”, as AI has a kind of constantly increasing creativity
reserved for humans and originally is an object; it can evolve through self-learning.
To some extent, such “intelligent creation” meets the basic condition to have “man”
through “labor” evolve into “civil man”, implying the possibility to rid AI of the
stereotyped concept of “object”. In this respect, when considering a person of civil
law, it should factor in AI, in particular as an author or creator.10 Some hold that
users of AI products should be entitled to relevant copyrights, which can incentivize
them to apply AI products and thus improve the development of AI technology.
Noncreator users of AI products should be regarded as fictional authors whose rights
can be decided in accordance with relevant right regulations in the Copyright Law
of the People’s Republic of China.11 In this way, the copyright of AI-generated
achievements is protected.
In patent rights, considering AI without the legal entity and the complexity of
relevant laws, legal policies are necessary, such as the designed system to intro-
duce AI as the “inventor” of technology, establishment of the patent standards for
AI-generated technical schemes, the rules to allocate rights and obligations of AI-
generated patent technology, and the mechanism for risk prevention and control of
AI-generated technology. These legal policies can resolve theoretical disputes and
barriers in practice. In this way, patent law can effectively regulate AI-generated
technical schemes.12 For copyright, the creation by AI learning and replication can
be copyrighted, as the work can reflect the unique selection and editing of compo-
nent elements by creators. Therefore, there is a risk of copyright infringement of such
creation. However, given the particularity of AI creation, vagueness can appear when
applying traditional material standards for copyright infringement. It is advocated
to introduce market substitution standards to assist the determination of copyright
infringement. A combination of substantial similarity and market substitution is the
best choice for the judgment on copyright infringement of AI creation.13

2.1.7 Tort Liability for Autonomous Driving in AI Applications

The convergence of different autonomous driving technologies comes along with

issues such as multiple actors, prolonged causation chains and ambiguity of the
liability boundary, causing three typical risk scenarios of autonomous driving: the
accountability vacuum in “a cart without the chauffeur”, the control dilemma in

10 See Liu Qiang. 2019. The Theoretical Challenges of Artificial Intelligence to Intellectual Property
Regime and its Responses. Rule of Law Forum (06).
11 See Xie Lin, Chen Wei. 2019. A Solution to the Dilemma of Copyright of Artificial Intelligence

Products under the Rules of Fictional Authors. Journal of Law Application (09).
12 See Liu Xin. 2019. On Patent Regulating of Technical Schemes Generated by Artificial Intelli-

gence—Theoretical disputes, Practical Problems and Legal Solutions. Science of Law(Journal of

Northwest University of Political Science and Law) (05).
13 See Zeng Tian. 2019. Research on Copyright Infringement Problem Created by Artificial

Intelligence. Hebei Law Science (10).

2 Theoretical Research on AI and the Rule of Law 49

human–machine cooperative driving and programs on the trolley problem. With

the further convergence and enhancement of autonomous driving technologies, the
autonomous driving system will take a core role in the traffic legal relationship,
and the legal liability will be shifted onto controllers of the system. It is neces-
sary to introduce the acceptable risk concept to ease the tension between the risk
of autonomous driving and innovation in science and technology, as well as the
improvement of the entire social welfare and the protection of individual interests.14
With regard to research on autonomous driving legislation, some argue that given that
autonomous driving technology and final autonomous driving vehicles—governed
by the law—are still under way, legislation on autonomous driving vehicles should
adhere to the ideas of innovation, vision and openness to strike a balance between
risk avoidance and innovation encouragement in a “progressive” way. In such a
“progressive” way, considering reality in China, one step ahead is to establish a
pilot autonomous driving area with separate legislation and administrative rules and
regulations, based on which we can pursue the ultimate goal of issuing the compre-
hensive Law on Autonomous Driving Vehicles. For legislative content, despite the
possibility that technological advancement may lead to various legal problems rele-
vant to autonomous driving, priority should be given to solve problems such as legal
status, legal liability and insurance, privacy protection, and network security in the
near future.15 In terms of the current hot issue of autonomous driving torts, some
propose that damage to people by intelligent vehicles is not regulated by the exis-
tent law in China, nor can relevant rules be analogized or applied to confirm an
individualized subject for responsibility. As the intelligence of autonomous driving
vehicles can incur the generation of tort liability, the legal relationship among the
drivers, manufacturers and developers of intelligent vehicles is different from that
of traditional vehicles. Hence, we should recognize the instrumental personality of
intelligent vehicles and properly handle the relationship between security and respon-
sibility. Intelligent vehicles should be classified into different levels, but the regu-
lations for all levels should be unified. We should distinguish autonomous driving
from human driving and place the instrumental personality of intelligent vehicles at
the center when designing relevant regulations.16

2.1.8 Supervision and Accountability in Blockchain Applications

Blockchain technology is experiencing rapid growth while posing many risks. Corre-
sponding regulation and supervision in China has brought some risks under control,

14 See Wang Ying. 2019. How is the Law Possible? Risk Scenarios of Autonomous Driving and Its
Legal Analysis. Law and Social Development (06).
15 See Li Shuo. 2019. Research on the legislation of self-driving cars. Administrative Law Review

(02).
16 See Xu Zhongyuan. 2019. On Legislation of Tort Liability for Autonomous Vehicles—A Focus

on Tool Person. Law Science (04).

50 Y. Cui

but deficiencies remain: the concept of the “blockchain without coins” is controver-
sial; the legal characterization of virtual currency is inadequate; ICO regulation and
supervision confronts a dilemma in reality; regulatory policies and existent laws are
not in harmony; the concept and regulation of “securities” have limitations; some
major concepts are not clarified in laws and regulations. Therefore, only one-size-
fits-all regulatory policies cannot work to deal with risks posed by the blockchain
in the long run. By reference to international experience, regulatory authorities in
China can exercise prudential supervision to open ICOs correctly while improving
the effectiveness and flexibility of regulatory means.17 Some argue that the clear
legal definition of the internal relation in the blockchain system is the basis for
supervision and accountability. However, the features of blockchain technology will
greatly impact the application of the theory and methods adopted in the establish-
ment of traditional legal relations. Moreover, it is hardly an easy job to generalize the
structure and legal nature of the internal relation in the system whose components
are of different shapes to serve a specific purpose. By virtue of the association and
difference between the contract and organization revealed by present theories, it can
disclose the “quasi-corporation” trait of the blockchain system. On this basis, we
should establish the supervision and accountability system with the platform at its
center and properly handle the relation among the supervision right, the judicial right
and the technical docking of the system. The blockchain system requiring permission
and approval should be regarded as a limited partnership, and obligations for each
part of the system should be properly assigned.18 Despite the neutral stance of the
blockchain technology itself, if actors use the technology for the purpose of illegal
crimes, this technology application can still be judged as hazardous acts in crim-
inal law. For example, to attract R&D investment or issue virtual currency illegally is
suspected of illegal fundraising; to swindle in the name of the blockchain technology
is the crime of fraud or contract fraud; to spread the blackmail virus for bitcoins is
suspected of the crime to destroy the computer system and fraud crime. It is worth
noting that the over the-counter transactions of virtual currency are not subject to
punishment, nor can illegal financing and issuance of virtual currency constitute the
crime of illegal business. In property tort crimes, the institution designated by the
central bank is suggested to price Bitcoin in reference to the foreign market price. If a
person covers up the committed crime, he is guilty of evading foreign exchange, not
the cover-up crime. In procedural law, there are different requirements in custody,
obtaining evidence and embezzlement of investigation. In particular, attention should
be given to obtaining evidence from electronic data and blockchain technology.19

17 See Deng Jianpeng. 2019. Dilemma and Solutions of the Blockchain’s Legal Regulation. Law
and Economy (03).
18 See Wang Qingsong. 2019. Defining the Nature of International Relationship of Block Chain

System and How to Allocate Liabilities. Law Science (05).

19 See Zhang Qingli. 2019. Illegal Risk of Blockchain Application and Criminal Response. Oriental

Law (03).
2 Theoretical Research on AI and the Rule of Law 51

2.1.9 Judicial Modernization with AI Applications

The development and application of AI has provided new approaches and means for
judicial modernization. Judicial data are resources of legal validity, but the problem
of “data isolation” is increasingly prominent. It is imperative to make progress in
regulation, principles and procedures for data sharing. By referring to the informa-
tion sharing platform to converge law enforcement and criminal justice and through
a coordinated combination of convergence and blockchain technology, the capacity
to integrate, exchange and share data should be enhanced to facilitate data exchanges
and sharing across departments, regions, levels and systems. In this way, the estab-
lished mechanism for the management and services of judicial big data can have
a standard operating procedure with full coverage in a mode to respond and share
data. On this basis, it can create an integrated sharing system of judicial big data
that combines macro guidance with micro practice, normal operation with emer-
gency response, and accurate statistics with scientific application in harmony, laying
the resource foundation for judicial information development.20 In contrast to tradi-
tional judicial informatization, the application of judicial big data and AI technology
presents the technical features of data preprocessing, algorithm dependence, self-
adaptation and domain specificity, as well as the features of full coverage, funda-
mental functions, key roles and openness in the information era. The fusion and
overlapping of the above features and the characteristics of judicial scenarios may
induce risks such as the dissolution of the inherent nature of justice, the weakening of
the subjective status of judges, the substitution of judicial reform objectives and the
loss of control of judicial reform results. To better prevent these risks, the application
of judicial big data and AI technology must stick to the ethical norms that set the
ultimate goal to safeguard the inherent nature of justice, root in enhancement of the
subject status of judges, function as an instrument and make innovation in a prudent
way driven by proportionality.21
In electronic court records, its reform plan is a cautious trial under the existing
litigation legal framework. This is a result of closing the gap between legislation
and judicial practice, not a fundamental answer to the application of information
technology in court records. The reform aims to optimize the method adopted in the
court record, not to change it fundamentally. Inspired by the court record mecha-
nism, the reform of electronic court records should give the jurisdiction a leading
place while taking into account the right to know, read, use and other litigation
rights of litigants.22 Online litigation is litigation that recreates legal procedures
in offline actions online through information technology. It is different from smart
courts and internet courts. The establishment of the online litigation system should

20 See Zhao Long & Liu Yanhong. 2019. A Practical Exploration of Integrative Sharing of Judicial
Big Data–Logical Construction from the Perspective of Breaking “Data Isolation”. Journal of Anhui
University(Philosophy and Social Sciences Edition) (06).
21 See Wang Lusheng. 2019. The Judicial Application of Big Data and Artificial Intelligence

Technologies and Its Ethical Regulation. Studies in Law and Business (02).
22 See Zhang Xingmei. 2019. Reflection and Construction of the Electronic Reform of the Court

Record. Law Science Magazine (01).

52 Y. Cui

uphold litigant-centric philosophy, enhance the binding force of honesty and trust,
and achieve functional equivalence of online and offline litigation. The application
of online litigation should be subject to the litigant’s option of the procedure and may
be influenced by subjects or litigations. In accordance with the litigant option, courts
should leverage the litigation command to regulate and guide the application of online
litigation within necessary limitations.23 In the application of big data for case trials
of the same type, the same-type cases confirmed with the same-type judgments is not
only the key requirement for judicial justice in China but also an important embodi-
ment of the important principle of equality before the law. With emerging AI and big
data, the traditional legal profession has been undoubtedly challenged and has gradu-
ally changed from passive to active. By integrating storage, integration, analysis and
decision-making, the application of big data in justice provides technical support
for the realization of same-type cases confirmed with the same-type judgments. In
response to the application dilemma of big data in judicial trials, a series of measures
can be taken, such as the guarantee of data openness and sharing, improvement of
the technology and regulations for its use, comprehensive consideration of factors
beyond law, algorithm transparency and reasonability.24

2.1.10 Summary of Current Research

AI development has brought changes to our lives with each passing day, during which
many new situations and problems appear. A great amount of research and discussion
has been conducted on the development of AI and rule of law. The summary of current
research comes as follows:
(1) The current research emphasizes establishing the rule of law system for AI in
legislation, law enforcement, justice and law abiding, as AI development cannot
proceed without a scientific rule of law system. Moreover, it is important to create
AI jurisprudence and set a scientific definition of relevant rights, obligations and
duties. It should build this system for AI to lay a solid jurisprudence foundation
to support the improvement and development of the system. It is stressed to
build a system of AI rules and regulations compatible with AI development.
Currently, the majority of AI applications are adopted in finance, transport,
medical treatment, and city construction. These sectors should take the lead
in forming scientific and effective industrial norms, devising corresponding
administrative rules and regulations and planning the local pilot areas. On the
basis of previous experience, a scientific and feasible legal system for AI should
be created in accordance with the legal procedure, guaranteeing AI development
under legal protection.

23 See Zhang Xingmei. 2019. The Conceptual Basis and Application Path of Online Litigation
System Construction. Tribune of Political Science and Law (05).
24 See Zhu Hui & Liu Chenhui. 2019. Research on the Application of Big Data in the Same Case

Trial. Journal of Law Application (20).

2 Theoretical Research on AI and the Rule of Law 53

(2) In the current research, it is emphasized to actively promote the application

of AI in administrative law enforcement, guarantee data openness and trans-
parency as well as algorithm justice and reasonability, improve the intelligence
of administrative decisions and the efficiency of administrative law enforcement,
increase the satisfaction of people with overall administrative law enforcement
and build the rule of law government. It should devise corresponding regulations
for administrative law enforcement, define the scope of authority, list rights, and
procedural steps for law enforcement agencies, ensure laws for law enforcement
agencies in AI relevant activities and constantly enhance the standardization of
administrative law enforcement. Attention has also been paid to the integration
of AI and justice. Accelerated efforts should be made in the development of
“data court and procuratorate” and “smart court and procuratorate” to achieve
the modernization of the judicial system and capacity.
(3) The current research focuses on the formulation of access standards for the
AI industry. Only the technology and products that meet standards and pass
the testing are qualified to be put into the market and circulation. It is held
to strengthen comprehensive management and control in networks, data, algo-
rithms, codes and other key areas and the ability to resist risks in design, applica-
tion, circulation, promotion and other key points of AI. A preemptive emergency
response plan should be drawn to address potential risks in domestic (inbound)
and foreign countries as well as people and machine ends. The joint mecha-
nism across sectors and multiple departments should be established to ensure
the prediction, prevention and control of potential risks under the rule of law
principle.
(4) The current studies emphasize the establishment and development of AI majors
in colleges and universities and vocational colleges. The academic commu-
nity believes that AI and interdisciplinary education should be added while
reducing those majors that tend to be impacted or even replaced by AI. It hopes
to improve the scale and quality of graduates from AI majors through digitalized
education, nurturing talent for AI development. The deep integration of AI and
law education is highly recommended, and greater efforts should be made to
develop cross-disciplinary talent with expertise in both legal knowledge and AI
technology.
(5) The current research emphasizes international cooperation in AI. It is a global
issue for all countries to further enhance legal guarantees for AI development
and ensure development in a healthy and ordered way. It is suggested to speed
up developing relevant international norms, promoting international cooperation
in the AI sector and joining hands to promote AI development and application.
The continuous development and deep integration of AI in all social sectors has
exercised significant and long-term effects on all aspects of economic devel-
opment and social life. Countries around the world have tightened their legal
supervision and control of AI to regulate and guide AI technology development.
Research on AI in China is of certain hysteresis, so it should keep an eye on new
problems AI generates in practice and constantly deepen theoretical research on
AI and the rule of law. In this way, it can promote AI technology development,
54 Y. Cui

improve the core competitiveness of technology, and strengthen exchanges and

cooperation with countries around the world together to build a community with
a shared future for humans.

2.2 Academic Monographs on AI and Rule of Law

An academic monograph is a theoretical work of writing about achievements of

scientific research in a particular academic area by the author. It is supposed to make
great contributions to discipline development while gaining international recogni-
tion. Compared with academic papers, monographs have more words and cost more
time in writing, and thus, the number of monographs is far less than that of papers.
This section describes monographs about AI and the rule of law published in 2019
in China, and the analysis of these works will be conducted from different dimen-
sions. Data samples herein come from bookstore websites, including Dangdang.com,
Taobao, Amazon.cn. With “artificial intelligence” and “rule of law” together as
keywords to search on these websites, 22 works about AI and rule of law that
were published in China from January 1 to December 31 in 2019 were collected.
Topics extend from “artificial intelligence”, including artificial intelligence, big data,
blockchain, unmanned driving, intelligent assistance, robots, informatization, being
smart, autonomous driving, autonomous recognition, knowledge integration, cloud
computing, knowledge graph, deep learning, machine recognition, smart court, smart
inspection, online service, human genome editing, being electronic; from “rule of
law”, including legal system, rule of law, regulations, laws, legislation, justice,
trials, judges, courts, procuratorates, law enforcement, administration, criminal, civil,
civil law, criminal law, administrative law, intellectual property, patent, constitution,
litigation, rights and obligations.

2.2.1 Themes

The monographs published in 2019 cover a wide range of both theoretical and
practical themes, such as AI, big data, cloud computing, gene editing, smart court,
criminal, civil and justice.
The Criminal Law in the AI Era by Liu Xianquan mainly introduces the criminal
risks posed by rapid AI development and notes the limitations and hysteresis of the
current criminal law system in solving criminal problems about AI with answers
to each of these questions. In Big Data, AI and People, He Yuan focuses on the
application of unmanned vehicles, the emergence of blockchain technology and a
series of other real situations. He elaborates on how lawyers treat AI from the legal
perspective, how laws deal with AI and whether it is necessary for China to complete
relevant legislation on robots as soon as possible. Justice System Reform and Prac-
tice and Exploration of Smart Court by Yang Linping summarizes the theoretical
exploration and practical plans of the people’s court in the scientific establishment of
2 Theoretical Research on AI and the Rule of Law 55

staff classification, optimization of trial assessors, improvement in procedures, and

judicial responsibility system; the replicable and transferable experience in people’s
assessors, family trials and other reforms; the brand new achievements in the deep
integration of justice with internet, AI and other modern technologies. It is of substan-
tial significance to the reform of the justice system and the research and practice of
developing smart courts. Among translated works in Chinese, Criminal Robots: Arti-
ficial Intelligence and Criminal Law by Gabriel Hallevy gives a clear answer to how
the existing criminal law system should respond to the damage caused by the AI
system in human society. Hallevy puts forward a comprehensive mature legal theory
of AI criminal responsibility and advises recognizing and selecting similar principles
from current criminal law. Hallevy proposes the specific thinking pattern of crim-
inal responsibility for various autonomous technologies in multiple situations. The
crime cases that may happen in real AI applications are listed, and the corresponding
response is made based on theories.

2.2.2 Publishing Houses

Among the publishing houses that published AI monographs in 2019, part of them
were university presses, while the rest were local or professional. All university
presses are managed by universities that boast rich education resources, such as
Peking University Press, Tsinghua University Press, China University of Political
Science and Law Press. These universities are highly sensitive to new trends and
topics, have excellent learning capacity and keep up with the times.
In the local and professional publishing houses, Law Press China published the
most monographs in 2019, a total of 7, or one-third of the 2019 publications. Shanghai
Renmin Chubanshe (spph.com.cn) followed closely with 6 publications in 2019. The
press is mainly located in Beijing and Shanghai, where the application of and research
on AI take a lead in China. In addition, their advanced economy and rich resources
facilitate AI development and applications.

2.2.3 Authors

Authors of these monographs, except for Chinese translation works, are renowned
jurists or jurists who set foot on the AI field earlier in China. One of them is Cui
Yadong, then Secretary of Party leadership Group and President of Shanghai High
People’s Court and current Secretary of Party leadership Group and President of
Shanghai Law Society. Of his two works published in 2019, Artificial Intelligence
and Judicial Modernization unfolds the “trial-centric” reform of the litigation system
with AI assistance in the Shanghai justice system; summarizes experience in the
R&D, application of the 206 System, as well as its major breakthroughs in business
and technology; and develops a detailed exploration on the theory of “AI and rule of
law” in the deep integration of AI and justice, the improvement of the justice system
reform, development of the AI and rule of law system, preemptive research on AI
56 Y. Cui

and rule of law, and advancement, regulation and guarantee of safe and reliable,
controllable, and sustainable development of AI. The Blue Book on AI and Rule of
Law in the World (2019), issued at the Rule of Law Forum of the 2019 World Artificial
Intelligence Conference, discusses and analyzes the latest policy planning as well as
laws and regulations on AI, research on AI and rule of law theories, AI application
in the rule of law and some typical cases related to the technology. Liu Xianquan,
the author of Criminal Law in the AI Era, is now the dean of Law School of East
China University of Politics and Law (ECUPL), the professor in criminal law of
ECUPL, the Ph.D. tutor, the leading scholar of criminal jurisprudence, which is the
key subject of Shanghai Municipal; the vice president of China Research Association
of Criminal Law, President of Shanghai Law Society of Criminal Law. He receives
special government allowances of the State Council and works on criminal law
research and education with outstanding contributions to AI criminal rule of law
research. Another author, Yue Caishen, is the professor and dean of the School of
Artificial Intelligence and Law of Southwest University of Political Science and Law,
one of the universities first to offer AI and law courses. Artificial Intelligence Law
Research edited by Yue in 2019 is a collection of papers under the topic of AI and
law, exactly in compliance with “AI and rule of law”.
Apart from natural persons, organizations have also published works. Thunisoft
Yuandian Legal Intelligence compiles A Legal Professional’s Guide to Artificial Intel-
ligence. This book lists the specific technique closely relevant to the legal industry.
From the perspective of legal professionals, it gives new interpretations of the tech-
nique elements, forming the underlying knowledge structure of AI accessible and
understandable to lawyers. It presents an exploration picture from multiple dimen-
sions by unfolding the integration and application of AI technology in the legal
industry over the past years. Thoughts on AI + Law Practice Chinalawinfo Digest
(2017–2018), published by Chinalawinfo.com, collects relevant papers published
from 2017 to 2018. It is divided into four parts: “Big Data Analysis”, “Internet +
AI”, “Topics in Focus”, and “Practice Discussion”. Compared with natural persons,
works by organizations are more visualized in the relevant structure design and the
presentation of summary.

2.3 Research Projects on AI and the Rule of Law

Research projects refer to a series of unique, complicated and interconnected activ-

ities of science and technology research with specific goals or purposes supposed
to be completed within the set time, budget, and resource limitations in accordance
with regulations.
Under the guidance of the National Medium- and Long-Term Program for Science
and Technology Development (2006–2020), research projects have been the most
significant indicator to evaluate the comprehensive strength and core competitive-
ness of universities and the research capability of individual scholars. An excellent
research subject must have a clear research topic of research significance, scientific
2 Theoretical Research on AI and the Rule of Law 57

methods and feasible plans to conduct the research. Based on the sources of research
budgets, research projects can be grouped into off-campus and on-campus projects.
Through the selection of research topics, samples are taken from off-campus national
projects supported by the National Social Science Fund of China (NSSFC) and law
projects of the Ministry of Justice, China Law Society and the Ministry of Education
to present the development of research projects in AI and rule of law.
It is of more analysis and reference value to put the national and provincial
projects at the column as budgets for these projects come from higher authorities
or the authorities in charge of the projects; the larger the budget, the greater impor-
tance the project is attached. As the CPC attaches greater importance to research
projects and the initiated projects cover a wider range of topics, the NSSFC, under
the guidance of the National Planning Office of Philosophical and Social Sciences,
has been an important channel to support fundamental research of humanistic and
social science in China. The NSSFC will finance universities and individual scholars
of excellent research conditions and capacities across the nation. The past years
have witnessed a surge of in-depth and valuable research achievements, putting
projects supported by the NSSFC at a guiding, authoritative and paradigmatic place.
Provincial projects supported by government and provincial departments in China
are included to increase the sample size. The selection of projects granted by the
Ministry of Justice, China Law Society and the Ministry of Education can make
samples more scientific and accurate.
In terms of keyword selection, this part adopts the concept of artificial general
intelligence to include more keywords. In addition to “artificial intelligence”,
extended keywords include “internet”, “big data”, “blockchain”, “unmanned driv-
ing”, and “cloud computing”. “algorithm”, “Internet+”, “electronic evidence obtain-
ing” and “information confirmation”.

2.3.1 Trend in the Number of AI and the Rule of Law Projects

According to the statistics of the NSSFC database, 343 law projects were launched
in 2019, 14 of which were research on AI and rule of law, accounting for 4.1%.
Compared with 2018, the number and percentage of projects on this topic both
decreased (Table 1).
Why do such projects decline? Since the project with the title of “artificial intel-
ligence” was approved in 2017, theoretical research on AI and rule of law has been
thriving. In July 2017, the State Council issued the Next Generation Artificial Intel-
ligence Development Plan as national theoretical guidance, further clarifying the

Table 1 Law projects by

Year 2018 2019
NSSFC in 2018 and 2019
Law projects 349 343
AI and rule of law projects 27 14
Percentages (%) 7.7 4.1
58 Y. Cui

Table 2 Provincial law

Year 2018 2019
projects in 2018 and 2019
Law projects 393 332
AI and rule of law Projects 39 13
Percentages (%) 9.92 9.94

strategic goals, tasks and approaches for new-generation AI development. Driven by

this, the academic theoretical research on AI and rule of law reached a small peak
in 2018 when research subjects appeared on applying AI in criminal law, evidence
law, product liability law and international law. In addition, the number of NSSFC
projects is limited. Despite a decrease in number, research topics in 2019 are still at
the cutting edge.
Among the 14 approved projects in the field of AI and rule of law in 2019, there
are 6 projects with the title of “AI”, 4 projects with the title of “big data”, 3 projects
with the title of “algorithm”, and only 1 project with the keyword of “blockchain”.
The search results of legal projects of “driverless” and “cloud computing” are 0.
According to the statistics of projects listed in notices of the Rule of Law and
Legal Research Projects, Law Research Projects, and Humanistic and Social Science
Research Projects by the Ministry of Justice, China Law Society and the Ministry
of Education, there were 332 provincial projects in 2019, 33 of which were research
projects on AI and rule of law, accounting for 9.94%. Different from the NFFSC
trend, the provincial projects on this topic, despite a small decrease, steadily occupy
approximately 10% on the whole (Table 2).
The topics of provincial and ministerial projects are varied. Through the keyword
search, in terms of headlines of these 33 projects, “AI (intelligence)” appears 10 times,
“big data (data)” up to 14, “blockchain” twice, and “autonomous driving”, “informa-
tion technology”, “internet+”, “algorithm”, “electronic evidence obtaining”, “digital
transaction”, and “information confirmation” once for each.

2.3.2 Levels of AI and the Rule of Law Projects

National projects supported by the NSSFC can be divided into the major, key,
general, youth and western levels. Provincial law projects include the major, key,
and general projects, as well as the youth research project/young and middle-aged
project, the fundamental research with prioritized incentives/commissioned project
and the project to support the Western region/special tasks. Each level requires its
own standards for the application of the project that receives the corresponding fund
and attention from the government.
AI and rule of law projects at the national level are mainly general projects and
youth projects. According to the statistics of law projects supported by the NSSFC
in 2019, there were 26 major projects, 216 general projects, 86 youth projects and
15 Western projects. In terms of AI and rule of law projects, there are zero major
projects, 8 general projects, 5 youth projects and 1 Western project, accounting for 0,
2 Theoretical Research on AI and the Rule of Law 59

3.7, 5.8 and 6.7% of the total at each level. It can be concluded that the youth project is
relatively easier to initiate. On the other hand, the absence of major projects indicates
that more attention should be paid to AI and rule of law research. As a cross discipline,
this research requires the leadership of senior researchers of comprehensive research
capability (Table 3).
Apart from the above reasons and the limited number of NSSFC projects, law
projects in AI confront obvious barriers to entry. In response, much room has been
reserved for incentives to support young researchers in provincial projects. According
to the statistics of provincial law projects supported by the China Law Society, the
Ministry of Justice, and the Ministry of Education in 2019, there were 2 major
projects, 15 key projects, 229 general projects, 5 youth research projects, 54 young
and middle-aged projects, 2 fundamental research projects with prioritized incen-
tives, 3 commissioned projects, 3 projects to support the Western region, and 19
special tasks. Nevertheless, the largest number of AI and rule of law projects goes to
general and young and middle-aged (or youth research projects) projects. According
to the statistics of AI and rule of law projects, 22 are general projects, accounting for
9.6% of the provincial law general projects and 6.6% of the provincial law projects;
8 young and middle-aged projects, 14.8 and 2.4%; 2 youth research projects, 40 and
0.6%; 1 special task, 5.3 and 0.3%.
Compared with provincial projects in 2018, there was an obvious absence in the
major projects, key projects, fundamental research projects with prioritized incen-
tives and commissioned projects of AI and rule of law in 2019. This situation is in
compliance with that in NSSFC projects, but on the other hand, it implies inade-
quate attention given to AI and rule of law research. One reason for it is the scale
of project levels. The major and key projects have a larger scale and receive more
funds. For key projects going into the significant research that have a good founda-
tion and accumulated experience or on the growth point of new disciplines under
the principles of the limited goals and scales with an orientation to highlights, the
fund is no less than 3 million yuan. Major projects usually comprise 4–5 key projects
of similar goals that are pulled together under the unified planning to be launched
one after another. With the major national strategic needs and the major scientific
frontier at the center, major projects provide scientific support for socioeconomic
development and national security. Research on AI and rule of law in China kicks off
at a later time, and the accumulation of research achievements has not reached a large
scale. Nevertheless, with the development of smart society, it is an inevitable trend to
launch the major and key projects of AI and rule of law. It is worth noting that young
and middle-aged projects and youth research projects can broaden the horizons of
young researchers in AI and rule of law research. Young scholars, exposed to a wider
range of fields, react more actively to current affairs and hot issues. In addition, the
barriers to the launch of young projects are relatively lower, enabling AI and rule of
law projects to occupy approximately 10% of law projects.
 60

Table 3 Percentages of AI and rule of law projects by NSSFC in 2018 and 2019
Year Major projects General projects Youth projects Western projects
Projects/Total Percentages Projects/Total Percentages (%) Projects/Total Percentages (%) Projects/Total Percentages (%)
2018 1/28 3.6% 17/219 7.8 7/87 8.0 2/15 13.3
2019 0/26 0 8/216 3.7 5/86 5.8 1/15 6.7
Y. Cui
2 Theoretical Research on AI and the Rule of Law 61

2.3.3 Institutes Conducting AI and the Rule of Law Research

Through the analysis and summary of the sample data of 2019 AI and rule of law
projects in universities collected from the NSSFC database and notices on provincial
law projects (by the Ministry of Justice, China Law Society, the Ministry of Educa-
tion), it can discover the following characteristics of the distribution of institutes to
do AI and rule of law research.

Increasingly Even Distribution

The regional distribution is relevant to socioeconomic conditions but, to a larger

extent, related to universities. Institutes conducting AI and rule of law research
projects in 2019 are distributed across China, including the Beijing-centered Beijing-
Tianjin-Hebei region in north China (32%), the Shanghai-centered Yangtze River
Delta in east China (23%), Hubei-represented central China (13%), the Chongqing-
centered southwest region (13%), Guangzhou-represented south China (13%), and
the Shanxi-represented northwest region (2%). North China ranks first with 15
projects, and East China ranks second with 11 projects. The two combined account
for the majority of projects.

Typical Advantages of Institutes

The national and provincial projects of AI and rule of law that Southwest University
of Political Science and Law applied in 2019 accumulated to a total of 5, China
University of Political Science and Law 3, Eastern China University of Political
Science and Law and Xiangtan University 2.

2.4 AI and the Rule of Law Research Forums

Academic forums are an important way to develop academic exchanges and theo-
retical exploration through reports or papers. The topics and quantity of forums
indirectly reflect the hot issues and development trends in this area. This section
will take “artificial intelligence”, “big data”, “blockchain”, “robots”, and “science
and technology” as keywords in the AI sector, together with “law science”, “laws”,
“rule of law”, “policy” and “education” as keywords in the law field. According to
the keyword search results, there were 60 forums on AI and rule of law research
held in China and abroad from January to December 2019, an increase of 10 times
compared to 50 forums in 2018. Due to the limitations of the methods, the retrieved
8 forums held in foreign countries are not exhaustive and thus will be spared from
the analysis. Hence, there are 52 samples in total.
62 Y. Cui

Fig. 1 Number of forums held in each month of 2019

2.4.1 Time and Quantity

The AI and rule of law forum was held more than twice each month in 2019. The
total number experienced phased growth with a half year as a cycle. Details come as
follows (Fig. 1).
The number of 2019 forums reached a small peak of the cycle in May and
December when it was at the end of the spring and fall semesters in universities.
Generally, 24 forums were held in the first half year, while the second half year
exceeded the first by 4 forums.

2.4.2 Places

Among 52 samples, 49 can be used for analysis because the places of 3 forums cannot
be confirmed. According to the statistics, forums are mainly held in first-tier cities,
such as Beijing, Shanghai, Guangzhou, and Hangzhou.
Among 49 forums, 12 are held in Beijing, accounting for 25%; 8 in Shanghai,
17%; 6 in Hangzhou, 12%; and 3 in Chongqing, 7%. The percentage of forums held
in other cities is less than 5%. It can be concluded that first-tier cities with relatively
higher economic development levels have a higher demand for and attach greater
importance to AI.
2 Theoretical Research on AI and the Rule of Law 63

Advanced technology is constantly improving the social and living environment

while bringing potential unknown risks in its applications. Justice and legal services
confront unprecedented challenges and development opportunities. As a leader of
developed regions in China, first-tier cities should accelerate the development of
AI hubs by holding forums to pool the wisdom of masters in academic circles and
industries. They also play a demonstrative role in how to follow up with the trend,
transcend the time, share the wellbeing of intelligence, respond to intelligence risks,
and make policies for intelligence while creating a good framework and foundation
for the development of the new-generation AI and rule of law.

2.4.3 Undertakers

Holding a forum usually involves multiple organizations, such as organizers, under-

takers and co-organizers (hereafter generalized as “undertakers”). As long as the
organization participates in the forum, it will be listed as an undertaker. Universities
are the major undertakers, and the government generally offers guidance. The partici-
pation and assistance of social sectors to hold forums are going up, including research
institutes, law societies, lawyer firms, funds, media, magazines and corporations
(Fig. 2).
(1) Universities are still the major force in holding forums. According to statis-
tics, of 52 forums hosted in China, universities as undertakers engaged in 29, over

Fig. 2 Number of forums held by undertakers

64 Y. Cui

half of the total. One reason is that experts and scholars are from universities, and
moreover, compared with other undertakers, universities with authorities on research
have more resources for research projects. (2) The participation of corporations in
forums has experienced a substantial increase. Baidu hosted the 2019 Forum on the
Development of Legal Services Industry under the theme of limited law and bound-
less AI; Bestone Inc. organized the 2019 Legal + Forum in December; the Legal
Technology Conference Forum was inaugurated by Thunisoft in September 2019.
(3) Most forums have been held by research institutes, firms/associations of lawyers
and law societies. National and local AI research institutes (research societies or
others in AI research) at all levels participated in up to 11 forums in 2019, closely
followed by firms/associations of lawyers and law societies with 9 and 7 forums for
each. (4) Government agencies have paid more attention. As many as 10 AI and rule
of law forums have been held by or under the guidance of government or govern-
ment agencies, substantial growth over the previous year. (5) Media and periodical
journals and magazines are newcomers in forum organizers. Intelligest hosted the
2019 Intelligest Forum with “Beyond the Law” as its theme. Together with Eastern
China University of Political Science and Law, Legal Vision held the Fourth Summit
on Emerging Legal Services Industry in December 2019.
Additionally, the Expert Advisory Committee and Forensic Expertise Center of
the government, as well as the UNESCO Higher Education Center, are involved.
From the above analysis, AI and rule of law research are mainly conducted by
colleges and universities and are gradually catching more attention from the govern-
ment and all sectors of society. The theoretical research in universities is gradually
aligning with the practical demands of AI and rule of law in society. Only when
the theory and social practical needs are connected can we move toward the exact
right direction in research and constantly push back the frontier to strive for the
development of AI and rule of law.

2.4.4 Themes

In recent years, forums have covered a wide range of themes. No longer limited
to ethics, it involves the separate sector law with both policies and the rule of law
discussed. In 2018, such themes were discussed as intellectual property, rule of law
development, law education, smart courts, robots, ethics, city, policy, economy and
laws and regulations. On this basis, 2019 forums stressed the combination of AI and
law education, the responsibility for the young generation and shared wellbeing of
the future. More attention has been given to forums on the application of AI in law,
apart from forums with the theme of laws regulating AI.
Against the backdrop of big data and 5G, “legal science”, “innovation in rule of
law”, “future of rule of law” and other keywords have been discussed in forums,
indicating the growing recognition of and attention to AI from society. In terms of
AI and law, the discussion has gone further from ethics, policies, legal technology,
legal risks and challenges, legal critical thinking, legal regulation and guarantees,
2 Theoretical Research on AI and the Rule of Law 65

and finally law education. Inspired by the research interests of experts and scholars,
it falls on responsibility for young lawyers.
Apart from the discussion on AI and rule of law in China, there are also in-depth
exchanges with foreign countries. In May, the Information Network and High-tech
Professional Committee of All China Lawyers Association and Guizhou Lawyers
Association held the China International Big Data Industry Expo 2019 that included
international forums, such as the International Forum on Regulations for the AI
Industry and Cross-Border Flows of Data, enhancing international exchanges and
cooperation in AI and rule of law. Currently, AI has already been a new type of
productivity in the world. AI and rule of law research can facilitate advanced tech-
nology to deliver more benefits to humans. Centralized discussion and exchanges at
forums can help people discover and resolve problems in AI and rule of law more
efficiently, speeding up the development of a theoretical system of AI and rule of
law and governance modernization in China.

2.5 AI and the Rule of Law Research Papers

Paper publication is a major indicator reflecting the academic research level in a

certain field, and it is also an important channel for the public to know the new
development trend in the relevant industries and fields. Hence, it is of great importance
to count and analyze the quantity and quality of the published papers of AI and rule
of law. Academic papers are a theoretical paper written after conducting research
on academic problems in a certain scientific subject. This part will analyze research
papers in China and beyond.

2.5.1 Data Collection

(1) Sources: Data samples of research papers herein are collected from CNKI and
CNKI Grid Resources Sharing Platform.25
(2) Date: The data were retrieved from January 1 to December 31, 2019.
(3) Method: It adopts the advanced search in the CNKI website. The collocation of
“keywords/topics + topics/abstracts” was applied to involve as many papers as
possible, and the research words were constantly updated to avoid data missions.
Words are grouped into two areas: “AI”, including artificial intelligence/AI, big
data, blockchain, unmanned driving, intelligence assistance, robots, informati-
zation, smart, autonomous driving, autonomous recognition, knowledge inte-
gration, cloud computing, algorithm, crawling/python, knowledge graph, deep

25 CNKI and CNKI Grid Resources Sharing Platform provides the richest knowledge information
resources and the most effective knowledge dissemination and digital learning to the high efficient
sharing of knowledge and resources in society by the self-developed and world-class digital library
technology, restoring “news”, “periodicals”, “theses and dissertations” and others.
66 Y. Cui

learning, machine recognition, 206 System, smart court, smart procuratorate,

online services, gene editing, digitalization; and “rule of law”, including legal
system, rule of law, regulations, laws, jurisprudence, legislation, judiciary, trial,
judge, court, procuratorate, law enforcement, administration, criminal, civil,
civil law, criminal law, administrative law, intellectual property, patent, consti-
tution, action, rights, power, and obligations. Given the various kinds of research
words, only part of them are presented here.
(4) Steps: It includes three steps: basic database establishment, correction and
verification.
Step one: The basic database was established by using “keywords/topics +
topics/abstracts” as key words to search the data, and finally, 2,862 valid data
items were left after the duplicate items were deleted.
Step two: In accordance with Price’s law26 and composite index,27 the database
includes 40 authors that make the greatest contribution to AI and rule of law
research. On this basis, published papers by these authors were retrieved from
the CNKI website by entering the author names in the advanced search, and
then based on retrieved results, correction was made to the basic database.
Step three: After the previous two steps, each piece in the enlarged database
was verified one by one, and the irrelevant pieces were deleted to make data
samples pure and accurate. The final data samples are 2,262 pieces.
(5) Data Processing:
The first is to confirm the regions of authors. The confirmation depends on the
institute location stated in the paper. To keep consistency in data analysis, for
the same author, if the institute or location in later published works is different
from that in previous ones, the institute location stated in the latest published
paper by the author takes prevail; if more than one institute is involved in the
same paper by one author, the first mentioned institute location takes prevail;
if one paper has more than one author, the institute location of the first author
takes prevail; the institute of authors are confirmed in the same way.
The second is to analyze citations and downloads of authors. One paper can
have more than one author. When calculating downloads and citations for each
author, the percentage value of the first author is 0.7, different from the second
author at 0.3. The rounded final result is drawn by multiplying the total by the
percentage value of each author. For example, as of the search date, Concept
Evidence and Form Orientation of Intelligence Rule of Law by Luo and Chen
had been downloaded and cited 685 and 4 times, respectively. By calculating,
for the first author Luo, this paper has been downloaded and cited 479 and 3
times, while for the second author Lei, it is one citation and 206 downloads.

26See Qiu Junping. 1988. Bibliometrics. Science and Technical Documentation Press. pp 188–190.
27See Qin Shoukang. 2003. Principle and Application of Comprehensive Evaluation. Publishing
House of Electronics Industry. p 16.
2 Theoretical Research on AI and the Rule of Law 67

The third is to define established journals in accordance with the eighth edition
of A Guide to the Core Journals of China in 2017.
The fourth is about possible data errors. Errors can occur in data collection, as
keywords in relevant areas cannot be listed exhaustively, resulting in errors in
collected samples. Additionally, in the preliminary data processing, the same
paper will repeatedly appear in rounds of searches with different subjects and
keywords. The repetition herein is defined as the duplicate of the same paper
title. It is resolved by clicking the button of “Remove Duplicates” in Excel.
Errors occur when different papers of the same title are deleted in this way.
The fifth is to prove the authority of data. Data are collected by a possible
exhaustive list of keywords in AI and rule of law to conduct unbiased searches.
During this process, noncompliant data will be excluded by reading abstracts
manually, ensuring the coverage of data as full and accurate as possible. To
prevent any data from omitting, after identifying the most contributive authors
by the composite index, their published papers will go through three rounds
of verification and then review. Relevant publications of these authors will be
added as well.

2.5.2 The Most Contributive Authors

Data Sources and Analysis

The bibliometric method usually needs to find the group of authors with the greatest
contributions in support of research in this field. Publications and citations are the
basic indicators in the author evaluation system. There is no significant positive
relationship between publications and citations: the great publications do not promise
great citations and while few in publications, they can be cited quite a few times.
If the publication is taken as the only standard for the author selection, it merely
presents the quantity but ignores the quality of papers. Hence, both publications and
citations should be indicators for a comprehensive evaluation of authors to reduce
errors. Price’s law is applied to identify the minimum publications and citations for
selecting author candidates. It will be selected as an example if the author meets
either the minimum number of publications or the minimum number of citations.
Here is the formula to calculate the publications: N1 = 0.749(Nmax)1/2 . N1 refers
to the minimum publication of a selected author; Nmax represents the maximum
publication of the author from January to December 2019. Only those authors whose
publications exceed the minimum can be selected as candidates. Among the authors
who collected 2,262 papers, Liu Xianquan had the maximum number of publications
with 9 papers, that is, Nmax was 9. According to the formula, N1 equals 2.25; that is,
those authors will be selected if their publications are no less than three. In accordance
with Price’s law, the minimum citation of published papers, N2, is 5.3; that is, authors
whose publications have been citied at least 6 times can be qualified. According to
the statistics, 25 authors have published over 3 papers, and 105 authors have their
68 Y. Cui

papers cited over 6 times. A total of 119 authors were selected as candidates after
eliminating duplicates of the same author.
First, it calculates the conversion index of publications and citations of selected
authors.
Publication Conversion Index = (publications of authors/average publications of
authors) × 100.
Citation Conversion Index = (citations of authors/average citations of authors) ×
100.
Second, it calculates the composite index of author candidates.
The publications and citations are weighted as 0.6 and 0.4 for each. The formula
to calculate the composite index is:
Candidate Author Composite Index = Publication Conversion Index × 0.6 +
Citation Conversion Index × 0.4.
Third, authors will be ranked from top down based on the composite index.
Fourth, authors whose composite index is no less than 100 (100 is the average
composite index) will be selected.

Descriptive Analysis

In the previous steps, 119 authors were selected as candidates. The average publica-
tions and citations for candidates are 1.76 and 8.92, respectively. On this basis, 40
authors that make the greatest contributions are selected.
According to the statistics, there are 40 authors whose composite index is over
100 in AI and rule of law, and they published 119 papers (the coauthored publication
is counted as one piece), accounting for 5.26% far less than 50%. To some extent,
this implies that the most influential authors in research on AI and rule of law have
not formed a group. In addition, researchers in relevant areas should go further and
extend the current research.
To further present the importance the leading law institutes have attached to AI, the
number of authors from “Wuyuan Sixi”28 and other universities is counted. Seven-
teen authors, nearly half of the total, come from “Wuyuan Sixi”, 5 from Eastern
China University of Political Science and Law, 4 from Southwest University of
Political Science and Law, 2 each from China University of Political Science and
Law, Renmin University of China Law School, and Wuhan University School of Law,
and 1 each from Peking University Law School and Jilin University School of Law.

28 “Wuyuan Sixi” is the abbreviation of five universities of political science and law as well as
law schools in four universities that were established after the founding of new China. Currently,
the legal subjects in these universities of great importance in the law education exercise profound
influence on the development of the legal system and rule of law in China. “Wuyuan” currently refer
to China University of Political Science and Law, Zhongnan University of Economics and Law,
Southwest University of Political Science and Law, Eastern China University of Political Science
and Law, and Northwest University of Political Science and Law. “Sixi” currently refer to Peking
University Law School, Renmin University of China Law School, Wuhan University School of Law
and Jilin University School of Law.
2 Theoretical Research on AI and the Rule of Law 69

The remaining 23 authors are from 18 other universities. The top goes to Southeast
University School of Law and Koguan School of Law, Shanghai Jiao Tong Univer-
sity, whose author quantity combined has exceeded six universities in Wuyuan Sixi.
This indicates that legal research at some universities tends to be at the forefront,
paying attention to hotspot issues of the rule of law. Over half of the authors are
from other universities, showing that universities have attached great importance to
AI and rule of law.

2.5.3 Research Journals

Distribution of Papers in Journals

There were 2,262 papers in AI and law literature databases in 2019. This part analyzes
the quantity of papers published in journals with more than 10 pieces. In 2019,
Legality Vision published the most papers up to 119 pieces, or 5.26%, a little less
than 150 pieces in 2018. It is followed by Legal System and Society, Oriental Law,
and Journal of Science, Technology and Law, whose publications are about the same
as the last year. Generally, 2,262 pieces are scattered in 854 journals, but only five
of them have over 20 relevant publications. A total of 429 (11.67%) papers were
published in these five journals, a significant 6.52% decrease compared with 2018.
Oriental Law and Journal of Science, Technology and Law, as core journals in China,
together published 50 pieces, or 2.21% in 2019, all about the same as 2.37% in 2018
that papers in the core journals with over 20 publications occupied.
According to A Guide to the Core Journals of China published by Peking Univer-
sity Library in 2017, of 2,262 papers, 548 papers were published in the core journals
in 2019, accounting for 24.22%, a 2.22% increase than the past year and 1,714
(75.77%) pieces were published in other journals. These figures indicate that the
core journals in China have given the same and full attention to research on AI and
rule of law both in 2018 and 2019.
548 papers are published in 171 core journals. According to Bradford’s law, the
formula is r = 2ln(e^ E * Y). Based on the AI literature database of this book, r repre-
sents the quantity of papers published in core journals, E refers to the constant,
0.5772 and Y is the maximum number of publications of the journal with the
largest number of publications, that is, 30. By calculation, r approximately equals
9; that is, the journals that have no less than 9 publications reach the core zone. In
this way, 19 journals are identified at the core zone, including Legal Vision, Legal
System and Society, Oriental Law, Journal of Science, Technology, E-Government,
Legal and Economy, People Rule of Law, Electronics Intellectual Property, ECUPL
Journal, Science of Law (Journal of Northwest University of Political Science and
Law), Modern Economic Information, Modern Law Science, People’s Procuratorial
Semimonthly, Regional Governance, Justice of China, Hebei Law Science, China
Publishing Journal, Youth Journalist, Police Science Research.
70 Y. Cui

CSSCI Law Journals

According to the latest CSSCI (2019–2020) source journal catalog, there are 24 law
journals that published up to 169 papers of AI and rule of law in 2019, accounting
for 7.5%. Each law journal has published 7 papers on average, which is significantly
higher than the average of 3 articles for other journals. This shows that law journals
of CSSCI attach greater importance to research on AI and rule of law.
Thirteen authors published more than one paper in CSSCI law journals in 2019.
The titles and research subjects of their papers will be analyzed. Liu Xianquan has the
most publications of 6, the major research subjects of which are “criminal law” and
“criminal liability”. Liu Yanhong also focuses on the criminal subject and includes
her. The following authors all published 2 papers. The subject of the highest frequency
that 13 authors research is the “criminal liability” for the severe behaviors of social
threatens occurring in AI and rule of law, about how to hold these behaviors account-
able, identify these liabilities and apply applicable laws, as well as how to make
interpretations and explanations of law. Much attention is paid to the torts of group
rights caused by the applications of AI, algorithms, big data and other technolo-
gies, ways to prevent these risks from happening and corresponding measures to
bring them subject to legal restrictions. Additionally, the application of AI in justice
and administration receives attention with the expectation of improving efficiency in
litigation and facilitating the implementation of justice reform.

2.5.4 Research Institutes

In terms of AI and rule of law research institutes in 2019, 78.37% (1,851) of them
were universities, taking an absolute dominating place. A total of 4.83% (141) are
government departments, including courts, procuratorates, bureaus of public secu-
rity and justice. Research institutes occupied 11%, higher than that of government
institutes, and party schools of CPC only accounted for 1%. These two combined
merely reach 12%, as the research subjects in these two institutes are rather distant
from those in AI, resulting in the small number of research papers. Five percent are
hospitals, magazines, newspapers, corporations and banks. Generally, universities
are the leading instates in conducting research.
The reason why two-thirds of research institutes are universities is that as AI is a
new emerging technology, its development of theory and practice is mainly driven
by universities to take a leading role in AI and law science research. In practice,
as AI has been widely applied in daily life, it is inevitable to deal with new judicial
cases involving AI. Procuratorates, courts and other justice departments will conduct
research on these new cases, so government departments will take up one-fifth of all
cases. Their research can offer new ideas and methods to handle many problems in
judicial practice while serving as guiding theory for the future development of rule
of law in AI (Fig. 3).
Universities accounted for the largest percentage in 2019, with a substantial
increase of 8% over the past eighteen years from 2001 to 2018. However, in 2019,
2 Theoretical Research on AI and the Rule of Law 71

Fig. 3 Institutes of AI and rule of law research in 2019 and 2001 to 2018

government departments experienced a decrease of over half, indicating inadequate

attention to AI research from administrative and judicial departments. Government
departments directly participate in law enforcement and particular judicial cases,
during which they have a chance to identify problems. Working on the frontline,
they should take the initiative in discussing relevant topics. Moreover, it is the call
of the theory and time to raise their attention. Different AI studies by government
departments can play varied roles in AI development in the legal area. AI research by
courts, procuratorates and public security bureaus can assist fair trials, prosecutions,
and prudent interrogations in AI-involved cases. The percentage of research insti-
tutes more than doubled compared with past years, showing that research institutes
in China have paid more attention to AI under the requirements of economic growth
and policies in the new era. Party schools of CPC have remained the same in recent
years.
In general, the national policies and the development of the national legal system
in 2019 both facilitate AI development with advancement in its theoretical and prac-
tical basis. The growth rate of universities is at a quite rapid pace, far beyond that of
other institutes. This swift growth has a close relationship with the development of
AI technology in reality. With the Internet and big data developing at high speed, AI
has gone from the illusion to reality in our daily life while exercising a significant
influence on people’s lives and work. In response, a theoretical analysis and research
on the influence of emerging AI development on law science is carried out. Due
72 Y. Cui

to the rapid development of AI, many unpredictable problems will appear in judi-
cial practice. Under the guidance of these problems, universities and other research
institutes will conduct corresponding theoretical research and innovation.
Following the trend in 2019, AI research by universities, government departments
and other research institutes in 2020 will exceed that in 2019 to reach a new peak.
It will make breakthroughs in AI and rule of law research, providing higher-level
guidance to AI-involved judicial practice.

2.5.5 Regions with Publications

Descriptive Analysis

According to the administrative divisions of China, there are 34 provincial-level

regions. According to the authors’ institutions, the regions are determined, and the
unmarked regions and foreign documents are eliminated. Then, 2177 papers are
sorted out. The regions are ranked in accordance with the number of papers published,
which can directly reflect the geographical distribution of AI and rule of law research
across the whole country.
The top three is Beijing, Shanghai and Jiangsu. Beijing has the obvious advantage
of 427 (19.61%) papers, a little less than 24.4% in the past years; Shanghai has
242 (11.11%) papers, about the same as 10.2% over the past years; Jiangsu is 149
(6.84%) pieces. Hubei, Guangdong, and Chongqing have 127, 120 and 112 pieces,
respectively. The median value is 50, and 12 regions have over 50 articles. It seems
smooth with little changes. Inner Mongolia, Ningxia, Tibet, Qinghai, Taiwan, Macao
and Hong Kong are still in the lowest ranking. In general, the 2019 research results
are similar to those in past years, and the size of AI and rule of law research is even
in most regions if the top three and the last seven are excluded.

The Eastern and Western Regions

There are 1,307 pieces in the eastern region of China, accounting for 61.38%, 338
in the western region, 15.99% and 469 in the middle region, 22.19%. The eastern
region has an absolute advantage, with 427 in Beijing, 242 in Shanghai and over
100 in Jiangsu, Guangdong, Chongqing, and Hubei. Fewer than 10 papers have been
published in Ningxia, Qinghai, Tibet and Inner Mongolia, requiring more effort.
Based on the above definition of the most contributive authors, it presents the
geographical distribution of authors. The eastern region has 33 people, while the
middle and western regions have 4 people each. Clearly, the eastern region has an
absolute advantage, particularly in Shanghai and Beijing. The gap is wide between
the eastern region and the middle and western regions, where increased attention
should be given to rule of law research in AI. Economy, culture and various other
factors cause such gaps.
2 Theoretical Research on AI and the Rule of Law 73

2.5.6 High Frequency Keywords

There are 32 keywords that appear more than 20 times. The keyword with the highest
frequency was “artificial intelligence”, 621 times, accounting for one-third of the
total. It is followed by “big data”, 580 times and “law”, 457 times. “Information”
and “the big data era” both appear 200 times. Then, the frequency of latter keywords
experiences a significant decrease, far less than the former ones.
Keywords in AI include artificial intelligence, big data, information, the big
data era, smart, blockchain, autonomous driving, algorithms, robots, and unmanned
driving. In AI-relevant justice, keywords include law, personal privacy, criminal
liability, copyright, intellectual property and privacy.
The 2019 rule of law research on AI is more concerned about legal restrictions
on personal information and protection of privacy, as well as accountability for tort
liabilities in unmanned driving and robots. Compared with the past years, some
major keywords of high frequencies do not change. Legal workers still focus on
two aspects: the emergence of AI will bring new problems in copyright, personal
information, legal liability and privacy; AI is applied in law, such as smart courts and
smart procuratorates.
Compared with 2019, the top ten keywords are artificial intelligence, big data,
big data era, personal information, blockchain, informatization, legal restrictions,
autonomous driving, internet and copyright (Fig. 4).

Fig. 4 Top 10 keywords in 2019 and before

74 Y. Cui

It is clear that the frequency of the top keywords is above the average level in AI
and rule of law before 2019, showing that more scholars focus on the core zone in AI
and rule of law, in particular the theoretical and practical discussion with “artificial
intelligence” and “big data” at its center. It attempts to treat the rule of law problems
in AI and other technologies in a holistic way, proposing theories for the concept and
subject status of AI. Some have discussed typical problems in AI practice, such as
autonomous driving and personal information protection. 2019 witnessed an increase
in these aspects and a substantial increase in more specific problems. This indicates
that the research in 2019 has a focus on specific problems, not only on principle and
abstract problems.
Compared with the past, more keywords were introduced to the legal area in 2019,
such as personal information, legal restrictions and copyright. This indicates that with
the rapid development of AI, increasing attention has been given to AI development
and its influence on legal theories and practices.

3 AI and Rule of Law Research Beyond China

AI and rule of law research papers in the world major countries in 2019 will be taken
as samples and collected from Web of Science.29 With “Robot” and “Law” together
as parallel subjects,30 526 relevant papers were retrieved and collected in 2019 from
Web of Science. The database of foreign papers is established on this basis. These
samples will be analyzed and then compared with those in China. In this way, it
hopes to present “AI and rule of law” research more objectively.

3.1 Research Subjects

The bar chart presents the top 20 subjects discussed in 526 papers. In general, foreign
scholars focus on the specific problems in AI application, while the focus in China is
on the theory. After analyzing the quantitative comparison chart of partial research
subjects and comparing the data with those before 2019, the following features of
AI and rule of law research in foreign countries can be discovered:

29 Web of Science is a unifying research tool for academic resources with a global coverage of
the citation index across disciplines. The website connects Web of Science Core Collection with
multiple regional citation indexes, patent data, indexing in a particular field and the research data
citation index. The multidisciplinary coverage of Web of Science encompasses 33,000 academic
journals. Relevant data provided by this database are of certain significance.
30 Based on previous experience of researchers, AI research in foreign countries has already changed

from the pure theory into application. Therefore, “Robot” instead of “Artificial Intelligence” is taken
as keyword for search. It turns out that with “Robot” as keyword, more data are retrieved, thus more
in line with the law of large numbers.
2 Theoretical Research on AI and the Rule of Law 75

(1) Foreign studies have focused on the technical aspects of AI rather than on
the jurisprudential aspects of the domain. The top 4 subjects are “Automation
Control System”, “Computer Science”, “Engineering” and “Robotics”, each
with over 350 publications, accounting for more than 65% of 526 papers in AI
and rule of law in 2019. Additionally, these are four classic topics in natural and
engineering science. In comparison, the top 4 subjects in China are “artificial
intelligence”, “big data”, “law” and “information”, each with over 350 publica-
tions, 621 for “artificial intelligence” and 580 for “big data”. Altogether, they
occupy 25%, far less than 65% in foreign countries. China focuses on theoret-
ical research, while foreign countries take AI technology as the subject. The
research on AI technology was 82.34% over the past 9 years, much higher than
that in 2019. This indicates that attention has been directed from the technology
to other aspects of AI.
(2) Foreign studies on AI and rule of law focus more on social science research.
Among the top 20 subjects in social science, it includes “MATHEMATICS”,
“BUSINESS ECONOMICS”, “BEHAVIORAL SCIENCES”, and “OPERA-
TIONS RESEARCH MANAGEMENT SCIENCE”. Compared with before,
the focus in the 2019 research is clearly directed to social science, while the
research subjects in social science remain similar in 2019 without significant
changes. This trend is natural, as the major foreign countries, in particular
Europe, America and other Western countries with the advanced rule of law,
have a long history of developing the legal system and the rule of law with
mature rule of law theories. The discussion will not be limited to the basic and
abstract topics of the rule of law and the legal system. Rather, in-depth research
based on the mature theories of these two topics will be conducted, such as
“business economics” and “operations research management science”. The top
research subjects in China are centered on the establishment of the legal system
and the rule of law, such as law, law science, criminal, justice, and legal restric-
tions. This is because against the backdrop of the primary stage of socialism,
the rule of law and the legal system are not well developed in China, which
is imperative to be improved both in theory and in practice. The difference in
research subjects between China and foreign countries reminds China of the
importance of developing the rule of law.

3.2 Authors

The analysis of the authors of the papers can better identify the groups of authors
who pay close attention to AI and the rule of law in the world. A total of 526 articles
were published in 2019, with a total of 2594 authors, which means that many of these
papers are coauthored, reflecting the features of AI. Artificial intelligence (AI) is a
new technological science that studies and develops theories, methods, technologies
and applications to simulate, extend and expand human intelligence. As a branch of
computer science, AI attempts to understand the essence of intelligence and create
76 Y. Cui

machine intelligence that can make reactions in a similar way as human intelligence.
The research subjects in this field include robotics, language recognition, image
recognition, natural language processing and expert systems. If the research in AI
and rule of law attempts to further study the impact of AI on the current or future
rule of law, it is necessary to have a certain understanding of AI itself in advance.
The edge of cross disciplines is thus highlighted.
In 2019, the most productive author with the most attention to AI and rule of law
research was WANG J, with 7 articles, accounting for 1.33%, followed by LIU L,
LIU LIU, ZHANG Y and ZHANG YM, each with 5 articles, 0.95%, and BEHERA
L, DAN WANG, HAO SUN and KIM J, each with 4, 0.76%.

3.3 Geographical Distribution

Following China, America, Iran, France, SOK, Canada, Britain, India and Australia
are at the top ranking in publications. Compared with before, Australia, India and
Iran jumped to the top for the first time in 2019. This shows that AI is a hot topic
that has received growing attention from more countries.
Currently, AI research is thriving. As a core technology, AI is studied in countries
around the world in full swing. China, the United States, France, Japan, Canada, Italy,
South Korea and Germany rank among the top publishers in the area of AI and rule
of law. The features of AI development itself include the following: AI technology,
based on the development of computers and neural networks, has demanding require-
ments for basic technology; there should be markets and scenarios that can have AI
applied in reality as only AI is applied in daily life can it work. AI development is
actually the advancement of technology, but technology advancement brings more
than technical innovation, and the application of technologies can cause fundamental
changes to all aspects in daily life. In particular, AI development sparks discussion
on the legal concept of robots in social science. The analysis of AI development in
each country can present the situation in that country while extending the perspective
from China to a broader world view.
In summary, most of the top ranked countries in the area of AI and rule of law are
developed countries. The large market and population, as well as the rapid economic
development achieved in recent years, offer realistic possibilities for the development
of AI and rule of law research in China. This is in compliance with AI and rule of
law development in China. Economic development has created a real demand and
resource for research in this area.
Chapter 3
AI Applications in the Rule of Law

Yadong Cui

1 Introduction

Applications of AI in rule of law hereof mainly refer to applying AI in legislation,

law enforcement, justice and notarization, legal services and other aspects. This part
analyzes the application in China and beyond from July 2019 to June 2020.
(1) Application of AI in legislation. In general, deep integration of legislation and AI
has not been achieved worldwide. The scope of adoption is relatively limited in
China, mainly centering on processes such as opinion solicitation and recording
and review. Due to the difference in regional economic development and tech-
nological advancement and other factors, this adoption diminishes from east to
west. AI-assisted legislation in foreign countries is still at the initial stage, with
few reports and research achievements.
(2) Application of AI in law enforcement. This application has reached a wide scope
with mature technology and significant progress worldwide. In China, it mainly
renders traffic enforcement, convenience services and criminal investigation,
transforming from postprocessing to preemptive warning and featuring clear
regional differences and prominent aggregation effects. Outside China, the main
scenarios take place in crime prevention, investigation and case solving and
border administration, but in facial recognition technology, generally speaking,
the EU and the US impose strict restrictions on its adoption.
(3) Application of AI in justice. The situation varies in each country. Great achieve-
ments have been made in China. With “smart jurisdiction” as the backbone,
through the top-level design and each-level implementation from top down,
real and significant effects can be seen. It is characterized by various kinds of
technologies, a broad coverage of applications in business, and a two-pronged
action of joint development of multiple actors and comprehensive and specific

Y. Cui (B)
Shanghai Law Society, Shanghai, China
e-mail: [email protected]

© Shanghai People’s Publishing House 2022 77

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2_3
78 Y. Cui

applications, and the quality, efficiency and credibility of justice has been promi-
nently improved. Despite differences in the levels of adopting AI in justice in
foreign countries, the general trend is to increase investment budgets for the
application with a focus on the R&D of the application system to predict cases
gradually.
(4) Application of AI in notarization. The adoption of blockchain technology in
notarization in China has achieved notable progress from evidence depositing
and obtaining to commissioned investigation and electronic notarizations, from
the data isolated island to the data integration, from the independent notarization
to the cooperation involving multiple parties, and from the single to the collab-
oration technology. Beyond China, active actions have been made in applying
the blockchain to the notarization. In addition to the establishment of the profes-
sional AI-empowered notary platform, it attempts to unveil more scenarios to
employ AI and other information technology in notarization.
(5) Application of AI in legal services. The application of AI in legal services in
China started late but is developing rapidly. The enriched AI-empowered legal
services can gradually provide different forms of services, such as O2O (online-
to-offline) platforms, online databases and document processing instruments,
to parties, law firms, courts and other service objects involved in the case. In
comparison, foreign countries developed legal services at an earlier time and
still possessed advantages in database building, algorithms and the underlying
technology, including the knowledge graph.

2 AI Applications of in the Rule of Law in China

2.1 AI Applications in Legislation

2.1.1 AI Applications at Different Levels

China has not yet achieved a deep integration of AI and legislation. Legislation is a
dynamic process covering bill sponsoring, study, drafting, deliberation, implemen-
tation and post evaluation of legislation, but to date, only several steps (i.e., opinion
solicitation and recording and review) have AI serve as an assistant. It is still at the
initial stage to make the whole legislative procedure informatized and smart.

National Level

The national unified R&R (recording and review) information platform has been
accomplished in China. Efforts have been made to connect the NPC R&R platform
and the State Council electronic recording system in an orderly manner. The informa-
tion platform lays the foundation for data sharing. Now, NPC is stepping up efforts
to build the database (phase one) of national laws and regulations.
3 AI Applications in the Rule of Law 79

Subnational Levels

There are multiple applications of AI in legislation at the subnational level, mainly

classified into two categories: one is the legislative information platform self-funded
by the government, such as the AI-supported legislative system of Shanghai Munic-
ipal People’s Congress and the AI-supported legislative information system of the
Standing Committee of SMPC, respectively responsible for collecting and picking
out revision suggestions for rules and regulations, and sponsoring, drafting and
reviewing bills; the other is the third-party AI-supported products of the legisla-
tive system procured by the government, such as the AI-empowered platform to
support legislation on PKULaw.com joint produced by Peking University Center for
Legal Information and Chinalawinfo Co., Ltd.
Compared with the national level, the subnational level renders more various
applications of AI in legislation. Only the information platform of Guangzhou
Smart People’s Congress has 12 applications, including the administrative system of
smart legislation. However, due to the absence of uniform standards for information
resource opening, much room has been reserved for improvement in this aspect.

2.1.2 AI Applications in Different Regions

Under the impact of regional economic and social development in China, the employ-
ment of AI in legislation witnesses a decline from the east to the middle and west
regions.
One application is the nationwide opinion solicitation system for draft laws.
Except for Hong Kong, Macao and Taiwan, 31 provincial administrative divisions
have created this channel to request public opinion. Nevertheless, the majority still
respond by letters, emails and faxes. Only 20 provinces provide online services to
submit opinions or leave a comment, 50% of which is located in the eastern region.
Despite adequate attention given to opinion solicitation at all levels, the standard
to apply AI technology in data analysis has not yet been clarified after collecting
these opinions.
This renders a noticeable difference in the adoption of AI-assisted products
in the eastern and western regions. For example, as of June 2020, a total of 15
provinces purchased legislative services supported by the AI-empowered platform
on PKULaw.com, and 8 or 54% purchased them in the eastern region. Additionally,
altogether, 5 justice departments/bureaus of provinces and municipalities purchased
service products of PKULaw.com, all in the eastern region. In comparison, the
demand for AI-supported legislation is not as high as that in the eastern region
during local governance in the western region.
80 Y. Cui

2.1.3 Exploration on AI Applications

New exploration of the recording and review process in legislation with AI assistance
has been developed on some platforms. In response to the demand of the legislature,
the AI-empowered platform to support legislation on PKULAW.com released a new
version in January 2020 in support of remote work online. It serves with intelligent
draft sponsoring, intelligent comparison, opinion solicitation for draft laws, recording
and review, version comparison, format generation and other functions. In recording
and reviewing, this intelligent platform has set up and gradually enriched the “sensi-
tive word library”, enhancing its learning ability. AI assists people’s congresses at all
levels in the screening and analysis of legislative opinions and post legislative eval-
uation, which is also the trend for future development of the legislative application
of AI. At present, it is rarely used in China.

2.1.4 Problems

Limitations of Services

Currently, legislative services with AI assistance are limited in the domestic market.
With PKULAW.com dominating the market of legislative service products devel-
oped by the third party, the services cover opinion solicitation, document review,
draft opening and other types of routine work. Technically, the product requires the
thinking pattern of legal persons to be translated into algorithms by the knowledge
graph, but the construction of the industrial knowledge graph is not mature. Generally,
smart legislation in China is still at the initial stage.

Less Practical Applications

Driven by the digital rule of law and the smart judicial information system initiative of
the Justice Department, the local legislature at all levels has made attempts to combine
AI and legislation. From the current open documents, the relevant practices are still
in the period of “introduction”, not “application”. The practice of “smart legislature”
at present suggests that legislation has a relatively high reliance on human resources.

Inadequate Theoretical Research

Currently, research on artificial intelligence and legislation in China mostly focuses

on regulating AI through legislation. On CNKI.net, “artificial legislation” and
“legislation” as search subjects were used to extract relevant papers.
While the research on AI goes to the highest, up to 34.16%, the research on
legislation is no more than 6.34%. If it is articles on AI-supported legislation, the
number is merely 3 to 5. Inadequate research on the theory of legislation with AI
3 AI Applications in the Rule of Law 81

support is not conducive to advancement in practical applications of AI technology

in legislation.

2.2 AI Applications in Law Enforcement

2.2.1 Regions of AI Applications

Given the complexity of public security work in different regions and the diffi-
culty in counting AI technology adopted in law enforcement, it searches and counts
the AI applications reported on official websites of public security departments in
31 provinces (except Taiwan, Hong Kong, and Macao) from July 2019 to June
2020, keywords including but not limited to “artificial intelligence”, “big data”,
“blockchain”, “facial recognition”, and “5G”. The basic database is thus established.

Clear Regional Difference

Through retrieval, the database comprises 107 accumulated applications that are
classified into the east,1 middle,2 and west3 regions for analysis and discussion.
The eastern region (54%) tops the rank, followed by the western (31%) and middle
(15%) regions. The peak of 11 applications appears in the east, while the trough
of 1 application appears in both the west and middle regions. This reflects clear
differences from the eastern, central and western regions.

Prominent Aggregation Effect

Through a respective analysis of the east, middle and west regions, it can be seen
that in the east region, the development is relatively advanced in Beijing, Shanghai,
Jiangsu, Zhejiang and Guangdong, altogether accounting for 78% of AI applications
in the region; Chongqing, Guizhou, Sichuan and Yunnan have a better performance
in the west, but the west falls behind the east in general; the percentage in the
middle region is split more balanced with Hubei and Anhui slightly higher than
others. Shanghai, Jiangsu and Zhejiang enjoy better development in the east, and
Sichuan, Chongqing and Guizhou top the rank in the west. This features the prominent
aggregation effect and benefits from the support of government policies such as

1 The east region includes Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian,
Shandong, Guangdong, Guanxi Zhuang Autonomous Region, Hainan.
2 The middle region includes Shanxi, Inner Mongolia, Jilin, Heilongjiang, Anhui, Jiangxi, Henan,

Hubei, Hunan.
3 The west region includes Sichuan, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Ningxia

Hui Autonomous Region, Qinghai, Xinjiang Uygur Autonomous Region.

82 Y. Cui

Fig. 1 Percentages of AI Hainan Shandong Fujian

Heibei
applications in the public 2% 7% 3%
2%
security of the East Region Guangxi
3% Zhejiang
Liaoning 19%
2%

Tianjin Beijing Jiangsu

3% 14% 17%

Shanghai
14%
Guangdong
14%

Fig. 2 Percentages of AI
applications in the public
security of the West Region

the integration of the Yangtze River Delta and the Sichuan-Chongqing-Guizhou

economic zone. According to the statistics, the ability of public security staff to
employ AI technology is closely related to the economic development level. If the
outlier is excluded, the majority of regions in China have no significant difference in
applying AI to law enforcement, which is connected with the long-term accumulation
of technology, investments and other factors. Shanghai, Jiangsu and Zhejiang and
Guangdong take a lead in economic development across China, so they remain more
agile in AI applications with a more solid foundation of basic technologies, such as
the internet. Beijing, as the national political center, has an increasing demand for
developing intelligent law enforcement of public security. Hence, these regions have
a better performance in the applications of AI in law enforcement (Figs. 1, 2 and 3).
3 AI Applications in the Rule of Law 83

Fig. 3 Percentages of AI Shanxi Hunan

applications in the public Inner 6% 6%
security of the Middle Mongolia
Region 6%
Hubei
23%

Jilin
12% Henan
12%
Heilongjiang Jiangxi
6% Anhui
17% 12%

2.2.2 Application Scenarios

According to the statistics, the application scenarios mainly involve big data aggre-
gation of public security, convenience services, criminal investigation, epidemic
prevention and control, case solving across departments, anti-drug business, stability
maintenance and anti-terrorism, traffic law enforcement, public security prevention
and control, financial supervision and prison management. Its various application
scenarios are connected with the multiple types of police officers in the public secu-
rity department. Among these applications, the top 3 goes to traffic law enforcement,
convenience services and public security prevention and control, accounting for 18%,
18% and 17%, respectively, altogether over 50%.
(1) In the scenario of traffic law enforcement, AI technology is mainly applied
in urban traffic brains, transportation, and travel services. The essence of this
technology is to connect and analyze information that the multitude of sensors

Fig. 4 Flow chart of traffic

brain4
84 Y. Cui

have amassed in each application scenario to exercise traffic management and

control in a centralized way (Fig. 4).
(2) In the scenario of convenience services, AI technology is mainly applied in
authentication, online customer services, information retrieval, intelligent self-
service terminals, and service robots. The level of applications changes from
perceptual to cognitive intelligence, which can be seen in the applications of
facial recognition, speech interaction, semantic understanding and intelligent
search. In general, cognitive intelligence is still at the initial stage, so there is
much room reserved for improvement.
(3) In the scenario of public security prevention and control, AI technology is
mainly centered on the continuous development of video surveillance tech-
nology, such as the earlier 311 demonstration project, Pingan Chengshi (City of
Peace), to the Xueliang and Tianwang projects. Currently, it has entered the stage
of intelligent surveillance when video structuring technology can autonomously
monitor, track and analyze the target, which is conducive to improving the effi-
ciency in solving cases. Facial recognition technology can match the face in
dynamic videos with the face library in real time, hence improving the accuracy
in tracking down a fugitive.
(4) In the scenario of epidemic control, AI technology has been widely applied
by the local government since the outbreak of COVID-19. The Public Secu-
rity Bureau of Shenzhen Municipality creates a complete closed loop of
“screening and warning, verification and home or centralized isolation” through
big data research and judgment by virtue of 13 intelligent application platforms,
playing a pivotal role in containing the spread of the virus. The employment
of big data analysis in epidemic prevention and control contributes to timely
informing of the epidemic and efficiently preventing the virus from spreading
and disseminating.

2.2.3 Types of Technologies

The application of AI in law enforcement covers 10 types of technologies. The top

4 are big data analysis (33%), facial recognition (28%), Internet of Things (9%)
and fingerprint identification (9%). The rest of them include vehicle identification
(8%), video structure (5%), gait recognition (3%), voiceprint identification (2%), AI
learning (2%), and iris recognition (1%).
(1) Big data analysis takes the largest share of the technology applications as the
foundation of applying AI to law enforcement is the building and analysis of
big data, while the recognition of faces and gait, as well as the identification
of fingerprints and voiceprints, is all based on analyzing massive data. On the
other hand, part of the system gives a vague description of their technology,
only ensuring that the basic technology is to analyze the big data, which is a
large base.

4 From iResearch, https://www.iresearch.com.cn/Detail/report?id=3396&isfree=0.

3 AI Applications in the Rule of Law 85

Fig. 5 Segmentation of the Government Industry, 2.7% Justice, 1.1%

big data intelligence market Work, 3%
Medical
of China in 20195
Treatment,
12.2%

Customer Finance, 38%

Services, 14.3%

Public
Security,
18.5% Others, 20%

The key of big data analysis lies in the construction of the knowledge graph,
the focus of which is on connections between concepts and entities, that is,
to interlink objects of complicated connections to maximize the degree of big
data analysis to the most. In 2019, the scale of the big data intelligence market
covering big data analysis and prediction and the knowledge graph reached
10.66 billion yuan. As public security occupies 18.5% of the market share,
ranking second, the R&D of big data analysis centered on knowledge graph
construction is the key for public security to employ AI technology. Generally,
AI applications in law enforcement have gained robust momentum and formed
a good ecology at the initial stage (Fig. 5).
(2) Facial recognition technology is relatively mature and highly integrated with
other technologies, with the second largest market share of 28%. The tech-
nology is mainly employed in convenience services, prison management and
criminal investigation, accounting for 56%, 17% and 10%, respectively. The
percentage of convenience services is the largest because the application of the
technology is at a low level with less difficulty. In prison management, it is
applied to monitor persons serving sentences, usually in conjunction with the
IoT. When employed in the investigation and evidence collection and the intelli-
gent arrest of the criminal investigation, facial recognition is usually combined
with the video structure and gait identification. The requirements for the tech-
nology in this area are very demanding, a must to reach dynamic recognition. It
usually integrates the video structure with facial recognition to process the video
contents automatically and match the face in the video with the one in the video
recording of the blacklist in real time to solve the case. For example, a man who
escaped and hid for 20 years was identified after his face was extracted by the
video surveillance system in the shopping mall and then matched with the faces
restored in the target information library. According to statistics, since public
security launched the Yunjian action in June 2016, the police with the help of

5 From iResearch, https://www.iresearch.com.cn/Detail/report?id=3553&isfree=0.

86 Y. Cui

the “portrait big data” system have captured 99,000 fugitives, a year-on-year
increase of 135.6%.6
(3) The IoT and fingerprint identification tied third place, each accounting for 9%.
The application scenarios of fingerprint identification are almost the same as
those of facial recognition, while the IoT appears more frequently in prison
management. For example, in a smart prison, handcuffs, wristbands, facial
recognition cameras and other IoT devices are applied to keep track of and
monitor the health conditions of persons serving the sentences. The moni-
toring center can receive the data collected by the control equipment and safety
equipment, enabling real-time safety warnings.

2.2.4 Development Features

From Data Isolation to Data Integration

Data have not yet been integrated or shared at a high level, impeding AI technology
development in public security. The data barrier is erected within the public security
system between different types of police officers, regions, levels and platforms but
also externally between public security and other departments. One of the features of
smart policing development is to continuously facilitate data sharing across levels,
regions, systems, departments and businesses from July 2019 to June 2020.
To ensure integration and sharing of information within public security, the
ministry, in accordance with the requirements of unified data resources, has created
a preliminary big data platform at the national level by the full data aggregation of
works performed by various police officers. At the subnational level, Guangdong,
under the planning to build the unified cloud platform of police, has completed the
provincial cloud resource service platform of police with eight innovations of police
applications in command, investigation, people’s living and prevention and control
and. The openness and sharing of data in public security with other departments
is also enhanced. In October 2019, the directory blockchain system across depart-
ments established by Beijing connected the responsibilities, directories and data of
53 departments in the city, resolving the absence and offside of data. The relationship
between public security and other departments and their workflow to share informa-
tion can be locked and chained by this system. The established rules for sharing data
can keep the data flow and business system in order.

From Single Actor to Cooperation in R&D

In AI technology development, with the needs of its work, public security fully taps
into the advantage of multiple actors and joins hands with high-tech enterprises,

6From the official website of the Ministry of the Public Security: https://www.mps.gov.cn/n22
53534/n2253535/c6840206/content.html.
3 AI Applications in the Rule of Law 87

colleges and universities and other government departments to develop public secu-
rity science and information technology research, application innovations and the
promotion of research achievements. Developers of 34 technologies have been clar-
ified. Even part of the technology involves dozens of developers. According to the
statistics, high-tech enterprises are the preferred partner with which public security
will develop cooperation with AI technology, accounting for 86%. For example, the
criminal investigation corps of the Beijing Municipal Public Security Bureau together
with 14 high-tech enterprises in China established a joint laboratory to design and
develop intelligent, scenario-specific and customized technology products that can
be applied in new-type cybercrime, traditional crime, professional crime and other
areas alike. The Hefei Public Security Bureau in Anhui cooperates with more than
ten renowned domestic companies and research institutes, including iFLYTEK and
HUAWEI. A series of new technologies, such as big data, cloud computing, IoT, AI
and 5G, have been adopted for public security.
Eight percent are colleges and universities who actively participate in the applica-
tion and development, and 6 percent are other government departments. In practice,
these two actors are usually joined to establish many AI laboratories, such as the
Sichuan Artificial Intelligence Research Institute coestablished by the government
of Yibin City, University of Electronic Science and Technology in China and Science
and Technology Department of Sichuan Province, and the innovation base of police-
civil integration by the Hefei Public Security Bureau and Institute of Advanced Tech-
nology, University of Science and Technology of China. This cooperation capitalizes
on the advantages of colleges and universities and government technical departments
to improve the efficiency of R&D and its application.

From Post-processing to Preemptive Warning

Relying on big data analysis and data mining, the support of AI to law enforce-
ment will turn from passive postprocessing to active and preemptive warning. The
employment of AI and big data has changed traditional retrospective investigations
to predictive investigations driven by big data. In the predictive model, the system
can automatically analyze the crime by connecting multiple factors of the space,
time, crowds, and crime types, with a conclusion of the hotspot, the high frequency
period and development trend of the crime and on this basis make a precise warning.
The mobile app for community policing designed by the Public Security Adminis-
tration Bureau of Guangdong Provincial Public Security Department is the smallest
unit of policing with an integration of “collection, judgment, actions and feedbacks”,
controlling the root of vicious incidents for prevention. From 2019 to January 2020,
preventable cases in Guangdong, such as robbery, snatch and theft crime, were
reduced by 21% year on year.
88 Y. Cui

2.2.5 Future Trend of Development

Focus on Ubiquitous Perception Plus Multimodal Data Integration

Ubiquitous perception refers to omnipresent perceptual intelligence, and multimodal

data integration aggregates the different types of data in voices, videos, characters,
images and other multiple information media. The 2019 market of perceptual intelli-
gence hardware devices is projected to exceed 35 billion yuan. The work to build the
public security view gallery based on the data for facial and vehicle recognition will
be accelerated to widen the boundary of the data structure and connect the business
and organs within the public security system. It is an inevitable trend to employ
ubiquitous perception and integrate multimodal data.

More Efficient Extraction of Data in Video Structure

The video structure extracts useful evidence accurately within a short period of time
after analyzing and processing the video content, playing a pivotal role in improving
the efficiency of case solving. Nevertheless, one video can contain complicated and
multidimensional information, and manual selection and definition of labels for each
feature are low in efficiency. Hence, future development is toward the highly efficient
extraction of video information. The extracted information should be interconnected
with data stored in other information platforms quickly to make unstructured video
information structured.

Continuous Advancement in the Combination of AI and Other Emerging

Technologies

In the 5G era, it is an irreversible trend to advance the deep integration of modern

technology and public security in society. Some regions have put that into practice.
For example, the Zhongshan Public Security Bureau in Guangdong has integrated
5G technology with AR technology, enabling the visual information captured by
front-end devices to be sent back to the command center in real time. “i Shenzhen”,
an officially launched app for unified government service, released the blockchain
electronic license application platform. The platform has realized 24 commonly used
electronic certificates, including the resident ID card, to be on the chain, ensuring
that each operation can be located.
3 AI Applications in the Rule of Law 89

2.3 AI Applications in Justice

2.3.1 Application Regions

AI technology is mainly applied in Smart Court and Smart Procuratorate initiatives.

These initiatives flourished across the nation from July 2019 to June 2020. In the
regional distribution, the application is centered on provinces with relatively devel-
oped economies, including Beijing, Shanghai, and Guangdong. Beijing continues to
optimize the Intelligent Trial system, the platform that integrates case sorting, online
mediation and the automatic generation of judgment documents and the application of
hot technologies in justice, such as the “balance chain” (Tianping Blockchain) access
technology. Shanghai, on the basis of the benchmark 206 System, made a further step
to develop high-quality systems, including the smart interpretation system, the intel-
ligent litigation service system for financial case trials, the smart preservation service
platform and the litigation risk assessment system. Guangdong also developed the
application of the smart credit ecosystem “Netcomchain” in 2019.
AI application in justice makes slow progress in the middle and west regions
due to the limitations of the economy and technology. In general, different degrees
of development to apply AI in justice have been gained, and the development has
advanced steadily in the 31 provinces in mainland China.

2.3.2 Hot Technologies

From the perspective of the applied technology, the technology of 2019 is supposed
to be blockchain and 5G technology.
Great achievements have been made in the integration of blockchain technology
and justice. The Beijing Internet Court “balance chain” and the judicial electronic
evidence perpetuation system based on the blockchain technology of the Hangzhou
Internet Court have garnered much attention from the public. From July 2019 to June
2020, the blockchain application in justice remains the center of evidence. The smart
preservation service platform of Shanghai No. 2 The Intermediate People’s Court
has data on the chain by blockchain technology, ensuring data preservation and real-
time retrieval. By leveraging blockchain technology, the Guangzhou Internet Court
created a trusted electronic evidence platform and “Netcomchain” in 2019. The most
significant feature of the Guangzhou Internet Court is the judicial credit report written
by the blockchain, forming the “five-color” system to evaluate the credit of internet
courts. Together with the judicial credit cogovernance platform and the intelligent
credit information evaluation mechanism, it provides new ideas for judicial credit
development.
With the pilot areas increasing, 5G has made improvements in the data transmis-
sion speed, virtual assistant performance, refined network trials and optimization and
90 Y. Cui

Fig. 6 Number of court live broadcasting trials in China9

upgrading of the end devices.7 The Guangzhou Internet Court has optimized online
litigation services comprehensively and is the first to conduct 5G-based online trials
in China. The rate of cases filed online reaches 99.94%, and the coverage and success
of electronic delivery are 99.97% and 98%, respectively. In July 2019, the trial was
broadcast live in Shenzhen Baoan District People’s Court by 5G technology.

2.3.3 Analysis of Applications

Online Trials and Livestream

On July 18, 2019, when the No. 426 case was live-streamed in Loudi Intermediate
People’s Court of Human, the number of trials broadcast online in China had broken
4 million, taking less than 3 years since China Trial Court Online was launched in
2016.8 According to the statistics of the website, courts that livestream their trials
have reached a large scale (Fig. 6).
Against the backdrop of epidemic control and prevention, online trials have
become a hotspot in intelligent justice applications. From January 1 to February 10,
2020, 286 courts opened the Smart Court in China, and 1234 trials were broadcast
live.
From February 3 to April 28, 2020, Shanghai High People’s Court conducted
13,173 online trials (mediation, hearing and conversations), over 270 times more
than 48 in 2019. Before 2019, the total was only 93. Similar online trial courts
include the “cloud trial court” system in Henan and Xinjiang and the livestream of

7 See Outlook for the Integration of 5G Network and Smart Court. WWW.CHINACOURT.ORG.
https://www.chinacourt.org/article/detail/2019/08/id/4297287.shtml.
8 From Smart Court: Contribution and Acceleration of Fairness and Justice.
WWW.CHINACOURT.ORG. https://www.chinacourt.org/article/detail/2019/09/id/4479470.
shtml.
9 From China Trial Court Online, retrieved from April 27, 2020.
3 AI Applications in the Rule of Law 91

the remote court trial in Shangxi. As of March 2020, over half of courts in China had
offered court trials online services, indicating the outbreak of online trials in courts
across the nation during epidemic prevention and control.10

Applications in Foreign-Related Judicial Activities

In 2019, some courts began to pay attention to the construction of AI systems for
foreign-related judicial activities. On December 6, 2019, Hainan Province No. 1
Intermediate People’s Court officially launched the first online diversified platform
to resolve foreign-related civil and commercial disputes. Hainan Free Trade Sport
contributes to the platform appearance. Its “one-stop” model to resolve disputes can
efficiently connect different platforms, including the TIAOJIE.GOV. CN and the
court trial system, enabling interconnectivity of data. On November 27, 2019, the
Shanghai Financial Court officially put into use the country’s first AI-based simulta-
neous interpretation system for court proceedings, which it developed independently.
The development of AI applications in foreign-related judicial activities provides new
application scenarios and ideas for the construction of AI judicial applications in the
future.

Public Interest Litigation Platform

Smart procuratorate development has gradually recognized the positive effect of big
data and AI technology on public interest litigation. Many procuratorates in China
have already developed on their own or joined with the private sector to develop AI
products in support of public interest litigation.
In September 2019, Fujian created the joint platform of the public interest litigation
and the procuratorate that can extract clues, sort out cases and share data in place of
manual labor based on the web crawler. In addition, it also designed a WeChat mini-
program. Together with the platform to obtain clues technically and the mini-program
flexible for the people to report clues, it makes the report of case information and
collection of clues in the public interest litigation more timely, flexible and accurate.

Mobile Micro Courts

Mobile microcourts, WeChat mini-program-based open online platforms for litiga-

tion services, make innovations in the online litigation of judicial trials by virtue
of social media software. As of October 2019, 31 provinces and Xinjing Production
and Construction Corps (XPCC) launched the mini-program of the provincial mobile
microcourt. Meanwhile, the mobile microcourt in each province has been connected

10From Remote Court Trials in Thousands of Courts in China, MZFZ.GOV.CN, https://baijiahao.

baidu.com/s?id=1660041231274073897&wfr=spider&for=pc.
92 Y. Cui

with the China Mobile Micro Court. By April 2020, more than 3,500 courts had
adopted the mobile microcourt in judicial practice. To date, this mini-program has
been a mobile electronic litigation service platform with unique characteristics.
Only in March 2020 did the China Mobile Micro Court accept 437,000 appli-
cations for filing cases online. A total of 361,100 of them are trial applications,
178,300 of which were filed. Among 75,900 executed applications, 37,400 cases
were filed. From January to March 2020, the highest number of applications appeared
in Zhejiang (94,956), followed by Henan (55,652), Shanghai (50,002) and Sichuan
(31,276), altogether accounting for 53.07%.11
In addition to applications for filing cases online, the China Mobile Micro Court
also adopts the online service of documents. As of May 20, 2020, up to 5 million
documents had been serviced online. Meanwhile, ten high people’s courts (Shanghai,
Zhejiang, Guangdong, Tianjin, Hubei, Jilin, Shaanxi, Guangxi, Qinghai, and XPCC)
conducted the remote court trial (the court trial online).

2.3.4 Features of Applications

From Independent to the Multi-actor Development

In 2019, developed AI judicial applications varied in types from AI-supported case

filing systems to intelligent court trial systems, rendering the features of multiple
actors involved. The absence of interdisciplinary talent is the core issue in the applica-
tion of R&D. Applications such as the “Four Smart Platform” in Gansu, the mediation
application of Qinghai Province People’s Court, systems of the full process online
case handling and intelligent case sorting in Shandong are developed independently.
The common feature they share is that the system architecture is relatively simple
with weak intelligence and few application functions. Currently, many AI compa-
nies have participated in the application of R&D, facilitating the development of AI
applications in justice in China.
Judicial AI applications codeveloped by judiciary actors and multiple actors have
powerful functions and can be innovated by deep integration with hot technology.
One typical example is the “206 System”, an achievement under the joint efforts
of Shanghai High People’s Court and iFLYTEK. The historical contribution of this
system is the introduction of the standard and guidelines for evidence to the system,
enriching the criminal theoretical and evidence systems. In addition, the Guangzhou
Internet Court has cooperated with over 30 companies, including three major mobile
carriers (China Mobile, China Telecom and China Unicom) and large internet compa-
nies, such as JD.com. Together, they developed “Netcomchain”, which attained good
testing results.
The existing practice has proven that only by combining the law of justice with the
cutting-edge technology of artificial intelligence technology can the application of

11From WeChat Official Account E-courts: Let Big Data Tell “Magic” of Micro Mobile Courts in
Convenience Services, https://mp.weixin.qq.com/s/0ZXxEPCbFPfJDT6JpE-AkA.
3 AI Applications in the Rule of Law 93

AI in justice move from the primary stage to the advanced stage, thus better serving
the judicial department.

Joint Development of Comprehensive and Specific Applications

The development levels of AI applications in justice vary with economic development

in each region, but judicial AI applications of comprehensive functions have been
promoted across the nation. In 2019, the intelligent judgment system “Xiaobaogong”
(xiaobaogong.com) was put into use or tested in over 200 courts and procuratorates
in China.12 The “206 System”, comprising four subsystems of public security, the
procuratorate, the court and the judicial administration, has been piloted in Anhui,
Shanxi, Yunnan, XPCC, and Guizhou. Capable of replication, promotion and exten-
sion, this system is a well beginning for the settlement of the application scenario
and geographic expansion of AI applications in justice.
When comprehensive applications are promoted with great efforts, the develop-
ment of specific applications also gains much attention. One significant achieve-
ment is the “balance chain” application of the Beijing Internet Court for evidence
collection and perpetuation. As of December 2019, over 10 million pieces of elec-
tronic evidence had been collected by the system and put on the chain.13 Shanghai
No. 2 People’s Court adopted the smart preservation service platform. This is the
first deep application of blockchain technology in property preservation in courts
across the nation. It includes four specific applications: guarantee letter verification,
preservation application, preservation change and rights relief.
Apart from evidence-specific applications, Shanghai High People’s Court deeply
integrated AI technology with trials in April 2019, thus deciding the best solution in
the application scenario of “the smart trial of financial cases”. The development of
specific applications can improve the application structure, functions and ability to
resolve target problems. In this way, AI applications in justice can be advanced with
real progress.

From Stage Application to Whole-Process Application

The AI application in justice has been extended from the early automatic generation
of documents and files to case filing and trials, almost covering the major steps of
the litigation procedure. Steps to exercise the service and supervision have received
attention in judicial AI applications.
To date, document service applications have been established in many regions.
Guangdong took the lead in building the service of short message applications. In

12 See Fan Chongyi, Several Questions about Sentencing Suggestions on Pleading Guilty and
Accepting Punishment, JCRB.com, July 15, 2019, p. 2.
13 From Beijing Internet Court: Over 10 m Electronic Evidence Deposited in “Balance Chain”,

Cover News, https://baijiahao.baidu.com/s?id=1651795112641416244&wfr=spider&for=pc.

94 Y. Cui

March 2019, Jiangxi created the provincial center for centralized service, inaugu-
rating the centralized and intelligent service model with service and trials separated
in China. Hunan also introduced an integrated postal delivery service platform that
can keep the record and give the timely feedback of the full information in sending
“court delivery” mails and receiving the “court delivery receipt”.
In procuratorates, the demand to apply AI in the supervision procedure appears
to keep up with the times. Procuratorates in Qingpu, Yangpu, and Minxing districts
of Shanghai started to pilot the integrated procedure supervision management plat-
form for criminal litigation and supervision. To ensure supervision throughout the
criminal litigation procedure, five data libraries are set up, including the compila-
tion of clues, illegal information, litigation supervision cases, similar-case supervi-
sion cases and comprehensive statistical analysis of data. Case filing, investigation,
review prosecution, trials and enforcement are covered to strengthen the weak link
in procuratorates.

2.3.5 Problems

Inadequate Investment

Inadequate investment has restricted the progress and promotion of AI applications

in justice. The ecosystem of AI R&D in China is not sound, and the court itself has
already been burdened with the fundamental research and organization of develop-
ment work, which has consumed considerable manpower and financial resources.
Hence, little can be reserved for AI application development. Even in metropolis
Shanghai, which has a developed economic level, insufficient funds have been a
bottleneck in technology R&D. Moreover, when AI-supported judicial systems are
finished, their later application and maintenance require a strong backup of funds.
The underinvestment problem has been a difficulty in applying AI to justice.

Data

(1) There are bottlenecks in building databases.

The AI application in justice needs data as the base for support. If the database is
incomplete, the application can hardly be conducted in practice. In addition to the
renowned PKULaw.com, courts are actively building their own databases. Shanghai
High People’s Court takes the lead in building the database based on data from
the three-level system of Shanghai courts, the portal information system and the
Government Online-Offline under the framework of Shanghai Municipal Big Data
Center.
The independent building of databases in each region leads to the data isolated
island phenomenon, making the data sharing within the country out of sight. As China
3 AI Applications in the Rule of Law 95

Fig. 7 Number and coverage of manual labeling cases in 2019

lacks guidance and planning for database construction, breakthroughs in judicial

databases can hardly be made any time soon.

(2) Generally, the data quality is relatively low.

Data quality is perpetuated in the whole process of applying AI in justice. Courts
themselves are in the position to generate data. During the period of data collection
and transformation, 10 courts in Shanghai adopted the single filing system. Only the
original nonelectronic paper documents, such as the IOU, need paper filing, and the
rest of them can be archived directly in electronic forms. This single filing system
can facilitate the full and effective collection of data and resolve the difficulty in data
transformation and recognition.
During the period of the data analysis, manual and automatic labeling based on
the intelligent standard generated from the manual standard was realized in Shanghai
courts. During 2019, 62,577 cases were labeled manually (the total manual labeling
accumulated to 122,577) (Fig. 7).
In automatic labeling, 16,572 cases of six crimes were tagged automatically. Auto-
matic labeling counted 694,296 times, accounting for 63.29% of the total (1,096,922)
(Fig. 8).
Nevertheless, the labeling work is a daunting task, as there is a large amount of
unlabeled data in the database and new data are generated at every moment.

(3) The technology stability should be improved.

In the questionnaire to survey the situation of AI applications in Rule of Law,14 as
for the question: how often do you encounter a bug in running the AI and rule of law

14This questionnaire surveys the practices of legal professionals to apply AI in the rule of law in
China. A total of 1,444 valid questionnaires were collected.
96 Y. Cui

Fig. 8 Number and coverage of automatic labeling cases in 2019

application software, 11% of the answer is “often”, 47% is “sometimes” and 42%
is “never up to now”. Bugs as answered in the questionnaire include: it is inflexible
to answer the question and often cannot understand the question; the case handling
system is a thankless task; the system crash; the wrong information.
Regarding the advantages and weaknesses of the smart trial system, most of them
think problems appear in the unified technical standard and the incompatible system,
low level of technology development, vague algorithm rules and other technical
aspects.
The core technical problems in AI and rule of law development remain unresolved.
Algorithms and technologies adaptable to justice are not mature and are unable
to adjust to its laws and characters. Few applications can truly fulfill intelligent
functions, such as assistance to sentencing of high precision.

2.3.6 Development Prospect

Auxiliary Position of AI in Justice

Some say that AI judicial applications will be the authority in place of judge. There
is widespread criticism of this argument. In the questionnaire, one question is about
the future of the intelligent trial system. A total of 310 people believe that the system
can provide a prejudgment reference to the case, and 363 people treat it as clerical
aids in case trials. In this sense, the frontline workers take the view that the intelligent
trial system can be an intelligent tool to assist judicial case solving, rather than the
AI product to replace human judges completely. It follows the development trend of
the AI rule of applications in the world to put it in this position.
3 AI Applications in the Rule of Law 97

Special Funds for AI Development

Despite much funds invested in AI development in China, the planning of special

funds for AI applications in justice falls behind. Special funds should be set up for the
development and maintenance of AI judicial applications. Much attention has been
given to R&D, while little attention has been given to practices. Under the backup of
funds, we can be free from this dilemma to make progress in parallel. Moreover, if
R&D facilitates practices and practices feeding R&D, it can form a virtuous circle.

Interdisciplinary Talents with Expertise in AI and Justice

Disciplinary talent training and reserving is a necessary step in AI judicial application

development. It requires disciplinary talent to fundamentally resolve technical depth
issues. For example, to provide powerful support for future AI judicial applications,
Southwest University of Political Science and Law and Shanghai University of Polit-
ical Science and Law established a school of AI and law, and Tsinghua University
School of Law established a computational law master program in 2018.

2.4 AI Applications in Notarization

2.4.1 Regional Distribution

The blockchain technology applied in notarization has gone from exploration and
research to application. In particular, advanced applications take place in Beijing and
Shandong of the Bohai Rim, Shanghai, Zhejiang and Jiangsu of the Yangtze River
Delta, and Guangdong of the Pearl River Delta. It is closely related to driving support
from the local internet court, enterprises and blockchain technology.

Drive of Internet Courts

Three internet courts in China all employ blockchain technology to create the
evidence platform: the “balance chain” electronic evidence platform in the Beijing
Internet Court, the judicial electronic evidence perpetuation system in the Hangzhou
Internet Court and the “Netcomchain” in the Guangzhou Internet Court. To enhance
the participation of notaries in auxiliary judicial affairs, the notary office tends to
join the blockchain platform established by internet courts. In this way, the notary
public office and the court litigation platform are connected efficiently.
From the cooperation between internet courts and notary public offices, Beijing
Changan Notary Public Office and Fangyuan Notary Public Office have joined the
“balance chain” electronic evidence platform of Beijing Internet Court as the first-
level node, enabling the notary public office to connect fully with the online litigation
98 Y. Cui

platform automatically; Guangzhou Nanfang Notary Public Office and Guangzhou

Notary Public Office also have joined the “Netcomchain” platform of Guangzhou
Internet Court. The ongoing pilot work of auxiliary judicial affairs with notarization
involved aims to efficiently connect notarization intelligence and court trials and
enforcement. To some extent, the application of blockchain technology in internet
courts contributes to the combination of notarization services and blockchain.

Impetus from Blockchain Enterprises

Blockchain enterprises provide the underlying technical support, powerfully

improving the applications of blockchain technology in notarization. These enter-
prises are mainly scattered in the Bohai Rim, the Yangtze River Delta and the
Pearl River Delta, where the levels of blockchain applications in notarization are
significantly higher than those in other regions. This implies that enterprises impede
blockchain applications in notarization.
In Jiangsu, “WuTongChain”, developed by Suzhou Tongji Blockchain Research
Institute and many other institutes together, is an alliance chain system platform
designed for exploring the application scenarios of the blockchain. It continues to
encourage more notary public offices to employ blockchain technology in scenarios
such as mask lotteries, degree notarization and deposit and evidence collection.

Support from the Blockchain Technology

The higher the blockchain technology development is, the fewer technical barriers for
the blockchain applied in notarization. In the regional distribution, the blockchain
application has a positive relationship with blockchain technology. It is the rapid
development of blockchain technology in Beijing, Shanghai, Hangzhou, Guangdong,
Jiangsu and Shandong that the blockchain applied to notarization in these cities
outnumbers others.

2.4.2 Application Scenarios

The blockchain is mainly applied in the notarization of electronic evidence deposit

and collection (23%), intellectual property notarization (27%), network notariza-
tion with mandatory enforcement (21%), lottery notarization (11%), blockchain
electronic notary certificates (9%), commissioned investigation (7%) and donation
notarization (2%).
The major application scenario is the protection and confirmation of intellec-
tual rights, accounting for 27%. Many notary public offices have set up specialized
intellectual property deposits and notarization platforms, such as the intellectual
property protection platform of the Hangzhou Internet Notary Public Office and
the “Hui Cun” blockchain electronic data storage platform of the Shanghai Xuhui
3 AI Applications in the Rule of Law 99

Notary Public Office. In addition, to solve the difficulty in electronic evidence depo-
sition, collection and confirmation and to make the electronic evidence chain fully
credible, each node notarized, and the whole process traceable, many notary public
offices, such as Beijing Zhengfang Notary Public Office and Shanghai Pudong Notary
Public Office, are exploring the one-stop electronic evidence notarization service,
including evidence collection, deposition, verification and cross-examination. Notary
public offices are also attempting to apply the blockchain to the blockchain notary
certificate, lottery notarization, network notarization with mandatory enforcement,
commissioned investigation and donation notarization.

2.4.3 Trend of Applications

Through the analysis of the blockchain applications in notarization, it renders the

following three major features:

From the Data Isolation to the Data Integration

Horizontally, the blockchain will connect data points into the “data strip”, forming the
blockchain notarization service across departments and regions. To promote cross-
departmental data integration, Beijing Shenan Weilai Technology Co., Ltd. with
the Beijing Zhongxin Notary Public Office colaunched the “Notary Alliance Chain”,
involving notary public offices, courts, banks, justices, big data companies and others
with a dedication to improving information integration among different departments.
Similarly, “Netcomchain” of Guangzhou Internet Court includes 8 units, including
Guangzhou Intermediate People’s Court, Guangzhou People’s Procuratorate, and
Guangzhou Notary Public Office.
The judicial alliance chain connects the data isolated islands in different regions
to promote cross-regional data integration. The Guangzhou Zhuhai Notary Public
Office, as the first judicial alliance chain notary public office node in the South-
east region, coordinates the electronic data directly with the Beijing Internet Court,
Guangzhou Internet Court and Hanghzou Internet Court through the judicial alliance
chain. Data can flow among regions.
Big data integration is the prerequisite for notarization intelligence. One of the
great trends of intelligent notarization in the future is to break the data isolated
island phenomenon and promote collaborative and cross-regional data sharing across
departments through big data integration (Fig. 9).

From Independent Notarization to Multiparty Cooperation

The integration of the blockchain and notarization drives the deep cooperation of
notary public offices with companies and research institutes who master the tech-
nology. As the blockchain itself is very technical, all notary public offices are no
100 Y. Cui

Fig. 9 Panarama of Notary Alliance Chain15

exception to cooperate with the third party. Technical support from a third party can
achieve the double credit enhancement of national credibility and technical trust.

From the Single to Collaborative Technology

Currently, notarization has a focus on blockchain technology, together integrating

IoT and 5G. The further integration of multiple technologies makes the notarization
intelligent and informatized.
A series of measures have been taken to regulate blockchain technology. From
the practices in notary public offices, the dominating trend is the alliance chain. It
can reduce the problems of the low validation of evidence and data isolation in the
private chain but also avoid risks of illegal and uncontrollable data contents in the
public chain.16
The IoT and 5G also penetrate notarization and integration with blockchain tech-
nology. Gandao Intelligence Technology Co., Ltd. released “Internet of Things Chain

15 From Ma Yiqing, Technical Solution Plans for Blockchain Notarization, Network Notarization
with Mandatory Enforcement, https://mp.weixin.qq.com/s/0z01VewrqoOxe5KtaBYCwQ.
16 Hangzhou Internet Notary Public Office: Support of No More Than One Errand, Notarization of

Law-Abiding Judicial Chain, https://mp.weixin.qq.com/s/oFGmM25KoQhi1Le4VM09Cg.

3 AI Applications in the Rule of Law 101

I”, the first IoT ecological blockchain communication control module in the world.
It has reached cooperation agreement with the Hangzhou Internet Notary Public
Office, realizing the combination of the blockchain and IoT. Hangzhou National
Notary Public Office and China Unicom in Zhejiang with Alipay organized the 5G
notarization technical team. The team designed the notarization system for online
room selection combined with the 5G high-definition video, resolving the video jam
in the traditional notarization video.
It is expected that 5G, blockchain and IoT technologies will be applied further in
the notarization. The application can empower notarization to improve its intelligence
and informatization.

2.4.4 Problems

Blockchain can solve the shortcomings of electronic evidence that is easy to damage
and tamper with, improve the cooperation ability of notaries across regions and fields
and expand business ecology and fields. However, there remain limitations. Details
come as follows:
(1) The ultimate trust problems cannot be addressed. The blockchain can only
ensure authenticity after it is put on the chain, while not before or during the
chaining, it is unable to keep online and offline data consistent.
(2) It challenges the “fully informed” principle in notarization. In traditional nota-
rization, the notary public office should fully inform the principal of his or her
rights, obligations, legal responsibilities and precautions and provide a clear
elaboration on the principal’s. In the intelligent internet era, it is completed by
the software system to inform the principal automatically in accordance with the
template. It remains doubt whether it can achieve the same effect and implement
the “fully informed” principle in traditional notarization.
(3) The uniform blockchain notarization platform is missing. Notary public offices
in different regions are exploring the independent combination of the blockchain
and notarization, forming multiple blockchain evidence deposit and collec-
tion platforms. A unified blockchain notarization platform has not yet been
established. To some extent, it prevented data from sharing and use across the
nation.

2.5 AI Applications in Legal Services

Lawyers in China are short in supply, and professional lawyers with master’s degrees
and above occupy a low percentage. The demand for lawyers is higher than the supply,
giving birth to the new form of services, “AI legal services”.
102 Y. Cui

Fig. 10 Applications of AI products in legal services17

2.5.1 Service Types

Currently, AI legal service products available on the market include O2O plat-
forms, online databases, robots for legal Q&A, physical robots, document processing
instruments and case prediction instruments (Fig. 10).

O2O Platform

As a matching tool to bridge lawyers and clients, AI technology applications on the

O2O platform are limited. The basic operation mode of the platform is as follows:
the lawyer creates an account; the client seeks advice; and the lawyer takes the order.
In the domestic market, there are two major types of platforms: the comprehensive
lawyer matching platform, such as Fa Xiaotao, and the domain-specific platform,
such as Zhiguoguo, the platform for intellectual property protections. Other forms
of O2O platforms in legal services are few in number.

Online Databases

China boasts the world’s largest judgment open platform, restoring massive judicial
data of high transparency. The development of legal big data products is usually
under the leadership of the government; one typical example is “Faxin”, a digital
network service platform for Chinese law applications.
The “Faxin” platform, initiated by the Supreme People’s Court and organized by
the People’s Court Publishing House, integrates big data and machine learning with
professional legal knowledge (graph) and then releases big data search engines, such

17 From EqualOcean Intelligence, 2018 Report on AI-Empowered Legal Services, https://www.

iyiou.com/intelligence/report551.html.
3 AI Applications in the Rule of Law 103

as “similar-case search” and “similar case for you”. In China, the similar-case search
engine is the only engine to provide a portrait of the whole case with the largest
number of dimensions and features.

Robots for Legal Q&A

Robots applied in legal services are customer service robots and the legal ques-
tion answering system. The former mainly renders in WeChat mini-programs, such
as “Xiaofa”, the official WeChat account of Xian Intermediate People’s Court.
The latter backed up by multiple AI technologies is highly professional. One
successful example in the commercialization of the legal question answering system
is “Fagougou”, which provides text and voice intelligent interactive solutions with
the support of the knowledge graph, semantic understanding, autonomous learning
and other technologies. Generally, in the domestic question answering market, the
customer service robots are large in number, but the answer quality is usually low with
little reference value, while the legal question answering system can provide advice
of high quality, but it is limited by the expensive development cost and the small
market retention. Given the limited number of lawyers in China unevenly distributed
(the majority in first- and second-tier cities) and expensive costs of legal services, it
is expected to have a strong demand for the mature legal question answering system,
and the market is of great development potential.

Physical Robots

Different from previous AI legal service products, physical robots render in the
form of terminals rather than online platforms or mobile applications. For example,
Shanghai High People’s Court launched the litigation service robot “Fabao”, meaning
the magic weapon of lawyers and courts. This robot and its WeChat counterpart are
implanted with the 12,368 knowledge base. In this way, they can meet people’s
demands for stereo search, legal advice and judge matching so that ordinary litigants
can be guided to get served effectively.18 Compared with communication terminal
applications, at least two shortcomings exist in physical robots: the price of physical
robots is so high that it reaches only a small number of people; physical robots are not
portable. Currently, physical terminal robots are mainly used in litigation service halls
of people’s courts at all levels with the expectation of improving litigation services.
They can provide services such as “litigation guidance”, “question answering” and
“legal aid”.

18New Upgrade in Shanghai “Smart Court”, New Litigation Service Robot on Duty, https://baijia
hao.baidu.com/s?id=1601706234450007195&wfr=spider&for=pc.
104 Y. Cui

Document Processing Instrument

Contract drafting and review are the major contents in AI-supported document
processing. Compared with human lawyers, AI legal service products with powerful
computing ability have a unique advantage in document processing. The lawbot,
rolled out by Shenzhen Lawbot Technology Co., Ltd., can complete 20 to 30 pages
of awards within one minute. However, it can only process contract documents with
a high percentage of structured data and high repetition. Many commercial contracts
have flexible expressions of terms and contain a large amount of unstructured data,
which are difficult to generalize in algorithmic or logistic languages. In this sense,
for companies, “AI prejudgment and manual review” is the best solution for human
and machine interaction.

Case Prejudgment Instrument

The case prejudgment of high precision is an important trend in the future develop-
ment of AI legal service products. There are mainly two types of operation modes to
prejudge cases: the data statistical analysis mode, which is similar to the statistical
analysis in the judgment of similar cases, and the winning party can be predicted
before litigation19 ; the rules embedded mode. In general, due to the lack of structured
data with labels, the prediction robots at home and abroad are still at the initial stage,
mainly centered on the prediction of simple elements, such as the winning rates of
lawyers, judgment results and years of sentence. Deep integration of the robots and
trials has been achieved.

2.5.2 Applied Technologies

The current legal service products present significant differences in employing AI

technology. Nearly all legal service providers claim their adoption of the underlying
technology, such as “big data” and “machine learning”. In fact, due to the unexplain-
able algorithm structure and business secrets, algorithmic information disclosed by
tech companies is not completely authentic. Comparatively, the O2O platform is at a
low level of intelligence, while the services of data retrieval, legal question answering
and document processing require the provider to ensure the real-time update of
the data warehouse and precise extraction of algorithms. In comparison with the
previous services, AI prediction services are uncertain and have higher demands for
the application of the underlying technology, including big data, logistic language
and machine learning. It has limited applications in practice.

19Four Prospects and Three Challenges in AI Legal Applications, http://www.chinalawlib.org.cn/

LunwenShow.aspx?CID=20081224141413590155&AID=20180330171318993289.
3 AI Applications in the Rule of Law 105

2.5.3 Development Features

Latecomer but Rapid Development

China is a latecomer in AI legal service products, but the iteration and release of new
products is in full swing. 2016 witnessed the launch of “Fa Xiaotao”, renowned as
the first legal robot in China; in 2017, “Longhua Xiaofa”, the country’s first legal aid
robot, was piloted; a year later, the country’s first humanoid legal robot, “Da Niu”,
made its debut at the 2019 Smart China Expo. Over the past three years, almost all
AI legal service products will experience one iteration each year in China.

Few Services and Weak Competition Mechanism

The government encourages the development of AI legal services, providing policy

support to the R&D of AI legal products. The basic platform of legal profession
knowledge and the data service “Faxin” facilitates the same judgments for similar
cases and regulations on sentencing in judicial ministries. In two consecutive years
of 2018 and 2019, “Faxin” was mentioned in the Report on the Work of the Supreme
People’s Court. Except for “Faxin”, the majority of products replicate only part of the
data restored on China Judgments Online, and few excellent legal services based on
big data have been rolled out by companies in the domestic market. The competition
mechanism for legal services should be further improved.

More Focus on Human Resources, Less on AI Applications

Currently, the legal service market in China still puts a focal point on human resources
with little attention on AI technology. AI legal service products have not garnered
many eyeballs in the domestic market. Few practical applications and upgrades in
AI legal products have been reported. In judicial practices, courts and other judicial
departments do not have a strong demand for AI legal products except for big data
platforms, such as China Judgments Online and “Faxin”.

2.5.4 Development Prospects

Emotion Algorithms

It is necessary to add an emotion algorithm to AI products to improve AI legal

services. If the legal robots are expected to master the negotiation skills and lead
the conversation, which originally is reserved for human beings, breakthroughs of
AI have to be made at least in “emotion recognition” and “answers with comfort”.
Currently, the Intel RealSense 3D camera has begun to microcapture facial expres-
sions of customers. The ability to comfort involves deep machine learning and the
106 Y. Cui

establishment of material libraries in different languages. A mature AI conversation

mode has not yet been established at home and abroad.

Expansion of Data Resources

AI legal services should be fed with massive data to obtain SOAT results. Although
the judicial open system in China takes a lead in the world, China Judgments Online,
China Court Trial Online, the archive of Chinese rules and regulations and the trial
management system in all courts cannot cover legislation and judicial interpreta-
tions at all local levels. In the future, AI legal service development should expand the
database capability while improving the openness of the data. Apart from legal knowl-
edge, it should add general data resources to break the limitations of the database
and help legal service robots develop individualized decision-making abilities.

3 AI Applications in Rule of Law Beyond China

3.1 AI Applications in Legislation

In the world, research and practices about AI-supported legislation are quite few in
number. AI application in legislation is still at the initial stage for preparation.

3.1.1 Theoretical Research

In the SpringerLink data, the keyword “legislation AND AI” was entered in the
search bar “with all of the words”, the resource type set “Article” and the time span
from 2019 to 2020. The results in the first-level discipline come as follows:
A total of 390 articles were collected (see Fig. 11). Research on AI legislation
is centralized on natural science, and little attention has been given to the law field.
This indicates that much room remains for improvements in AI legislation research
in foreign countries.

3.1.2 Practical Applications

Online Voting System

The National Assembly of the Republic of Korea and State Duma launched a
blockchain-based online voting system in 2018 and 2019, respectively. In theory,
the advantages of decentralization, immutability, and anonymity possessed by the
blockchain are conducive to ensuring security in the voting process and improving
3 AI Applications in the Rule of Law 107

Fig. 11 Retrieval results of “legislation AND AI”

the trust of the people in the final results. Nevertheless, blockchain technology is
not mature in the practical application of legislation. In June 2020, one observation
node in the blockchain-based voting system to help decide the amendments to the
Constitution of the Russian Federation was hacked. It is pressing to improve the
technical security guarantee and mature encryption measures in AI applications in
legislation.

System for Identifying Conflicting, Redundant and Outdated Regulations

In 2020, the AI tool RegExplorer developed Deloitte, which helped Ohio identify
conflicting, redundant and outdated regulations. This tool can help legislators review
legislative documents efficiently and effectively by providing research results in the
context.

Video Conferencing System

In April 2020, the House of Lords announced that “oral questions, urgent ques-
tions, ministerial statements and nonlegislative debates”, all of which were happening
entirely over Microsoft Teams, would be live streamed. Later, due to technical and
data compliance issues, the virtual sittings were declared not to be broadcast live.
In May 2020, the House of Representatives announced that senators, members who
were in isolation, or affected by COVID-19 for other reasons can make a remote vote
through their agent. Given the conferencing budget and the security system, video
conferencing has not been widely applied. Nevertheless, as COVID-19 seems no end
within sight, this technology is worth advertising.
108 Y. Cui

Text Analysis System

Japan’s first ordinance to tackle gaming addiction passed by Kagawa Prefecture was
enacted in April 2020. According to the Kagawa Prefectural Assembly on March
17, 2020, this ordinance received a total of 2,686 public opinions, including 2,269
“yes”, 401 “nay” and 16 “suggestions”. The text mining method is applied to analyze
opinions on legislation. The adopted text recognition technology developed by LINE
BRAIN can read and digitize public comments, model the dependency between lines,
extract keywords and save as an Excel sheet. Then, public opinions can be grouped
and analyzed.

3.2 AI Applications in Law Enforcement

3.2.1 Focus on Police Service in the Technical Aspect

The AI application in law enforcement unfolds based on the practical needs and appli-
cation of policing work in crime prevention, investigation and case solving, border
administration, stability maintenance and anti-terrorism and convenience services.
(1) Crime prevention. In July 2019, the Tokyo Metropolitan Police Department
together with Panasonic Corporation was working on a new kind of crowd
prediction system. The crowd data captured by the camera set on the police car
will be uploaded to the cloud for computing and analysis to predict the future
movement of the crowd. The real-time prediction system can detect suspicious
situations, flag out unnatural or suspicious activities and direct the police to
proceed with investigation. In the same period of time, Manila and Davao cities
in the Philippines purchased 12,000 CCTV cameras, aiming to aid the Philippine
police in the fight against crimes through facial recognition.
(2) Investigation and case solving. Facial recognition is widely applied in this area.
In August 2019, the British police adopted facial recognition software on smart-
phones to solve cases for the first time. It is capable of analyzing the photo of
the suspect and matching it with the criminal database in real time. In this way,
the suspect can be identified. In October, the Investigative Committee of Russia
made a statement claiming that 3D facial modeling technology was success-
fully developed, even to identify those with masks, the main purpose of which
is to help spot any lawbreaker. In December, the Kolkata police station in India
expanded the coverage of AI-driven video surveillance equipment of facial and
gait recognition functions for criminal investigation. In October, the New Delhi
government invited tenders to build the largest self-proclaimed facial recog-
nition system in the world. The new facial recognition system can centralize
faces data captured by scattered surveillance cameras in India and connect the
data with the national database to aid the Indian police in discerning criminals,
missing persons and bodies.
3 AI Applications in the Rule of Law 109

(3) Border administration. In February 2020, George Boltadakis, the coordinator

of the European Dynamics Luxembourg project, said that iBorderCtrl funded
by the EU is designed to reduce time law-abiding non-Europeans spend on
land border checkpoints. In addition, polygraph technology, including auto-
matic spoofing detection, biometric recognition, facial matching and document
authentication, is further employed to help border guards identify illegal immi-
gration and prevent crimes and terrorism. This system will be piloted in land
crossings of Greece, Latvia and Hungary.
(4) Stability maintenance and anti-terrorism. Intelligent robots have been widely
applied to execute dangerous tasks, such as investigation and EOD. In September
2019, the US put the X-ray detector “HBI-120” into trial, which could discover
explosives and other contraband goods hidden in cars or behind walls. In
November, the powerful “dog-like” robot, called Spot, was put into use in
America. As a mobile remote observation device, it can generate images of
potentially dangerous places. Since January 2020, the first batch of Guardian
XO in beta version has been delivered to the US military for trials. Guardian
XO, as the first wearable robot powered by batteries in the world, can safely lift
a weight of approximately 100 kg.
(5) Convenience services. On November 8, 2019, the multimodal biometric national
ID card system was launched in the Republic of Angola on the west coast of
Southern Africa. This system collects citizens’ fingerprints, portraits and other
biological characters for information collection, extracts features and verifies
them.

3.2.2 Focus on Privacy Protection in Theory

With the growingly mature technology, in theory, it is widely debated whether AI

technology can continue to be applied in law enforcement, in particular facial recog-
nition technology. The major argument is whether the privacy of citizens can be
protected.
(1) US. States of the US take a cautious attitude toward it, believing that as the adop-
tion of facial recognition technology will infringe on the privacy of citizens and
undermine their rights and freedom, the scope of its application is generally
restricted. California State Assembly passed Assembly Bill 2261 about facial
recognition technology or California Facial Recognition Bill on February 14,
2020. It prohibits any organization from applying the facial recognition service
for continuous tracking and surveillance, otherwise in specific situations of
enforcement investigation on serious criminal offences. The House of Repre-
sentatives and Senate passed Senate Bill 6280, that is, Washington Facial Recog-
nition Bill on March 12, 2020. It is provided that the state and local government
can employ the facial recognition service to locate or identify missing persons
and determine the identity of the deceased, including the missing or murdered
the female Indians, those fit for Amber Alert and Silver Alert and other possible
victims, to ensure public security. Later, San Francisco proposed the draft of
110 Y. Cui

Ethical Use of Facial Recognition Act. It prohibits government organizations

from installing any equipment related to facial recognition technology and from
obtaining individual information by accessing or using facial recognition tech-
nology. California State Senate also rolled out a bill to ban the police from the
use of the full body camera with facial recognition or other biotechnology. The
Somerville City Council in Massachusetts voted to pass a bill, which prohibits
the adoption of data or evidence generated by the facial recognition software
system in any criminal investigation or litigation. US Democratic Senators Jeff
Merkley and Corey Booker proposed the Ethical Use of Face Recognition Act
on February 12, 2020. It requires any government organization to be prohib-
ited from employing facial recognition technology until the usage guidelines
and restrictions are prepared by the committee to prevent facial recognition
technology from violating citizen rights protected in the first amendment of
the US Constitution, that is, the infringement of privacy and freedom. On June
10, 2020, the US tech giant Amazon posted a short statement, announcing that
the US police was banned from its facial recognition technology for one year.
According to the comparative study by MIT, Amazon’s facial recognition has a
higher false positive rate when identifying faces of dark color, possibly leading
to increased risks for the police to misidentify people of color and increased
errors in law enforcement. Perhaps, that is a possible reason for the ban.
(2) EU. In terms of personal privacy protection, the EU is recognized as the region
with the strictest restrictions in the world. In May 2018, the General Data
Protection Regulation was enacted by the EU, regulating the use of biometric
data, including facial data. In December 2019, the EU Commission said it was
considering implementing the most stringent regulatory measures on AI. In the
draft of White Paper on Artificial Intelligence to be released, if the public or
private organizations employ facial recognition technology in the public place,
a 3- to 5-year ban will be imposed. After full consideration, the ban on facial
recognition technology was removed, but strict restrictions were imposed on
the remote biological identification system. In November 2019, the European
Union Agency for Fundamental Rights released the report of Facial Recog-
nition Technology: Fundamental Rights Considerations in the Context of Law
Enforcement, with a focus on the fundamental rights involved in facial recogni-
tion. CNIL in France published Facial Recognition: for a Debate Living up to
the Challenges on November 15, 2019, clarifying the concept of facial recogni-
tion and assessing relevant risks so that it can reject or shoulder with appropriate
guarantee some risks based on democratic decision making.
3 AI Applications in the Rule of Law 111

3.3 AI Applications in Justice

3.3.1 Significant Difference in Regional Development

In the general development of AI and rule of law applications worldwide, information

about the popularity and application of judicial AI in foreign countries is obviously
short, mainly because courts are slow to embrace new technologies in justice.20 Due
to the notable difference in economic levels and varied importance attached to AI
development in each country, the development of AI judicial applications renders
significant differences. Several years earlier, the US was already familiar with risk
assessment software, such as COMPAS and LSI-R, and many state courts were
able to employ algorithmic software to aid the judge in sentencing. However, in
the 2019 upgrade of File, the electronic document filing system adopted in Oakland
County, the obvious change remained the change in rules for use, rather than the
upgrade of the system.21 Similarly, in California courts, the SRL web portal devel-
oped to improve the delivery of online services to self-represented litigants (SRLs)
as well as all other court users throughout California is under design and deployment
yet with little progress.22 Germany has been trying to build a specific AI system
to crack down a certain crime, such as the to-be-developed AI system against the
child pornography crime. AI judicial applications in Canada have made considerable
progress. Mediation robots based on AI algorithms can replace manual mediation
successfully.23
There is a huge difference in regions with undeveloped economic levels. For
example, it was not until January 2020 that local courts in Malaysia were ready to
introduce AI court technology to assist sentencing.24

3.3.2 Case Prediction

With AI technology developing over time, big-data-based AI case prediction systems

are playing an increasingly important role in judicial AI applications worldwide.
Judgment prediction applications based on big data and AI have been built around
the world, such as ROSS and Lex Machina. Nevertheless, few technical innovations
were reported, except for revisions of some usage rules from July 2019 to June 2020.

20 See Peter Suciu: AI in the Courts: The Jury Is Out, Tech News World, https://www.technewsw
orld.com/story/86521.html, Last Accessed Times: May, 16, 2020.
21 See Oakiand Circuit Court: eFiling, https://www.oakgov.com/courts/circuit/programs/Pages/eFi

ling.aspx. Last Accessed Times: May, 15, 2020.

22 See California courts: California Courts Digital Services and Self Help Redesign, https://www.

courts.ca.gov/42412.htm, Last Accessed Times: May, 15, 2020.

23 From Sifa Sina, AI: Perspective of Intelligent Mediation Negotiation System, Sina, http://news.

sina.com.cn/sf/news/fzrd/2019-08-28/doc-ihytcitn2550925.shtml.
24 From Chen Zhihong, Panarama of Future Courts Unfolding in Malaysia, Tecent, https://new.qq.

com/omn/20200424/20200424A0°C8800.html.
112 Y. Cui

Case prediction for the assistance of case solving is projected to be one of the focuses
in future AI judicial applications.

3.4 AI Applications in Notarization

Foreign countries are actively employing emerging technologies such as blockchain

technology in notarization. A professional blockchain-based notarization plat-
form has been built, but the application scenarios of the emerging technology in
notarization are also being expanded.

3.4.1 Blockchain Notarization Platform Expansion

In foreign countries, there are already many blockchain deposit and notarization
platforms, such as Factom, Stampery, Custos, SilentNotary and Blocknotary, dedi-
cated to improving trustworthy confirmation and notarization services based on the
blockchain and creating a more open and transparent environment for confirmation.
In July 2019, Blocknotary released the blockchain electronic diary for the conve-
nience of staff to add, check and tag notarization behaviors. The encrypted notariza-
tion behaviors are under the protection of the blockchain in the prevention of edit and
delete actions of outsiders.25 On September 26, 2019, the SilentNotary development
ushered a new stage. In the face of anonymity and the broken causation chain, Silent-
Notary created a centralized and distributed registry platform expected to be applied
in company businesses.26 In May 2020, EOS launched a new notarization service—
proof of file existence from blockchain—to prove the existence and ownership of
files.27

3.4.2 Expanded Application Scenarios

On September 1, 2019, Brazil issued the first birth certificate recorded exclusively by
using blockchain technology. The registration through the Notary Ledger platform
was made much faster.28 On November 11, 2019, the first civil partnership recorded
on the blockchain took place in Brazil through the distributed Notary Ledgers network
of local startup Growth Tech, the first time when the blockchain was leveraged for

25 Notary and blockchain—the perfect combination? https://bitnewstoday.com/news/notary-and-

blockchain-the-perfect-combination/.
26 A new stage in the development of the Silent Notary project! https://silentnotary.com/#news.
27 EOS Launched New Notarization Service: Proof of File Existence from Blockchain, https://www.

chainnews.com/articles/896645056144.htm.
28 First Blockchain-Exclusive Birth Certificate in Brazil, https://www.qkljw.com/article/5983.html.
3 AI Applications in the Rule of Law 113

civil partnership notarization.29 It is worth noting that on June 1, 2020, Court of Rio
de Janeiro announced that blockchain was not allowed to “conduct electronic notary
acts”, which blocked the blockchain technology applied to notarization.30
In August, 2019, PUBLISH Inc. launched the beta version of an “onchain” news
article notarization service in Korea. By clicking on the hyperlinked hash value,
readers are able to check the full text of the article on the blockchain and key
information, conducive to cracking down on mounting fake information around the
world.31
In 2020, Italy also employed the blockchain for diploma notarization, able
to directly check background information of students in prevention of certificate
fabrication effectively.32

3.5 AI Applications in Legal Services

AI applications in legal services have cover wider field at deeper levels from July,
2019 to June, 2020.

3.5.1 Community Groups and Service Export

On October 21, 2019, Ross Intelligence announced reaching deep cooperation with
Clio, the renowned legal cloud service provider. The AI-driven legal research plat-
form ROSS, funded by the world’s largest law firm Dentons, was made open to Clio
users, achieving great breakthroughs in user community groups and coverage.33 Clio
once secured 250 million dollars in funding, the largest in legal technology history.
In addition to cooperation with relevant companies, the team also reached agree-
ments with the Florida State Bar Association and Illinois State Bar Association in
June 2019 and February 2020, respectively. ROSS will provide more comprehensive
services to lawyers and their assistants in these two associations. Under the guid-
ance of the Law School Program, ROSS gives law students across the US access to
its industry-leading AI legal research platform.34 Kira Systems, the AI platform for

29 Brazil’s First Civil Partnership Recorded on Blockchain, https://www.360kuai.com/pc/9e9779

408d6ac5d14?cota=3&kuai_so=1&sign=360_57c3bbd1&refer_scene=so_1.
30 Rio court does not recognize blockchain use by notaries, https://bitcoindynamic.com/news/rio-

court-does-not-recognize-blockchain-use-by-notaries/.
31 PUBLISH launches blockchain-based content notarization service in beta——TokenPost,

https://coinhubnews.com/blockchain/publish-launches-blockchain-based-content-notarization-ser
vice-in-beta-tokenpost/.
32 Italy: Blockchain in Diploma Notarization, http://m.xiaobaicj.cn/p32170.html.
33 Clio and ROSS Intelligence Join Forces to Redefine Legal Research, https://blog.rossintelligence.

com/post/clio-and-ross-intelligence-join-forces-to-redefine-legal-research.
34 ROSS Intelligence Launches Law School Program, https://blog.rossintelligence.com/post/ross-

intelligence-launches-law-school-program.
114 Y. Cui

assistance in contract review, subsequently joined hands with Litera Microsystems35

and Bryan Cave Leighton Paisner.36 It can to a great extent improve work efficiency
and reduce the risks of errors and omission in legal work, enabling us to focus on
the analysis and suggestions of high value.

3.5.2 Continuous Optimization of AI Platforms

On April 8, 2020, ROSS Intelligence announced the availability of the complete

Statutes & Regulations for all 50 States, the United States Code, and the Code of
Federal Regulations on its A.I.-powered legal research platform. It is convenient to
access legal information in all jurisdictions.37 Another AI platform, eBrevia, inte-
grated with iManage, enables users to seamlessly leverage its AI-based contract
analytics solution in tandem with iManage’s secure document and email manage-
ment system and brings unprecedented accuracy and speed to contract analysis due
diligence, knowledge management and compliance-related tasks.38

3.5.3 AI Legal Services Thriving During the COVID-19

The worldwide spread of COVID-19 challenged the legal service industry at large.
Teleworking and online offices dominated the work mode during this special time.
In this aspect, AI has played an unparalleled role that the traditional work network
does not. For example, Kira offered a one-time document limit allowance to users
free of charge to better meet the greater demand of document review and improve
the efficiency and accuracy of review.39

35 Litera Microsystems Announces Kira Systems as First Content Provider in Litera Connect
Program, https://kirasystems.com/blog/litera-announces-kira-connect-program/.
36 Bryan Cave Leighton Paisner Makes Strategic Global Commitment To Embed Kira Within High-

Volume Workstream, https://kirasystems.com/blog/bclp-embeds-kira/.

37 ROSS Adds Statutes & Regulations For All 50 States, https://blog.rossintelligence.com/post/

ross-adds-statutes-regulations-for-all-50-states.
38 eBrevia Integrates with iManage to Extend Document Management Capabilities for Legal

Teams, https://investor.dfinsolutions.com/press-releases/news-details/2019/eBrevia-Integrates-
with-iManage-to-Extend-Document-Management-Capabilities-for-Legal-Teams/default.aspx.
39 How Kira Systems is Helping Our Customers Respond to the COVID-19 Pandemic, https://kir

asystems.com/blog/helping-our-customers-respond-to-the-covid-19-pandemic/.
Chapter 4
Al-Related Case Analysis (China)

Yadong Cui

1 Overview

In recent years, Al has penetrated every aspect of people’s lives and profoundly
affected our lives and production. In March 2016, AlphaGo shocked the world by
defeating Lee Sedol, a top South Korean Go player of 9 dan rank. The event ushered
in a new era of AI’s rapid development. It has become one of the most cutting-edge
technologies and the hottest topic.
To acquire an in-depth understanding of disputes over AI-related cases, we
collected 569 copies of AI-related judgments, among which 371 copies were dated
from July 2019 to June 2020 (Data sources: China Judgments Online, Wolters
Kluwer).
We provide a visualized presentation of time, geography, litigants and case types
by analyzing AI-related cases filed in recent years and provide an overview of AI-
related cases in Chinese judicial practice to support scientific legislation and the
application of law.

2 Time Series Analysis of AI-Related Cases

2.1 Explosive Growth of AI-Related Cases

We analyzed the dates of AI-related judgments. As is shown in l distribution shown

as Fig. 1.

Y. Cui (B)
Shanghai Law Society, Shanghai, China
e-mail: [email protected]

© Shanghai People’s Publishing House 2022 115

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2_4
116 Y. Cui

Fig.1 Distribution of AI-related Cases in Recent Years

Figure 1 shows that since the State Council issued the New-generation Al Devel-
opment Plan in 2017, the widespread use of AI technology has brought about a
significant increase in the number of AI-related cases.

2.2 AI-Related Cases Are Concentrated in the Last Three

Years

Since 2016, the number of AI-related cases has experienced exponential year-over
year growth of 36, 71 and 248, which implies that including modern technologies
such as AI in national strategies has significantly promoted its development. At
present, AI has truly entered people’s lives, and a rule of law system is urgently
needed to ensure its regulated and orderly development.

3 Analysis of Geographical Distribution of AI-Related

Cases

The application of Al technology varies in different regions due to their perspective

on social economic development. We mapped out the geographical distribution of
AI-related cases (see Fig. 2) by analyzing the court’s locations of 569 judgments,
from which we can estimate the application of AI in different regions.
4 Al-Related Case Analysis (China) 117

Fig.2 Distribution of Litigants in AI-related Cases

3.1 AI-Related Cases Were Concentrated in Western

and Central China

Table 1 shows that in terms of the number of AI-related cases, Beijing and Guang-
dong provinces exceeded 100, accounting for 40.2%; Hubei, Jiangsu provinces and
Shanghai municipality exceeded 20, accounting for 29.5%; Zhejiang, Hunan and
Shandong provinces exceeded 20, accounting for 14.4%. These provinces and cities
have relatively high levels of economic growth, especially Beijing, Shanghai and
Guangdong, which are regarded as the center of China’s economy. Meanwhile, these
provinces and cities adopt more optimistic policies on technology industries. They
have a great number of high-tech companies and higher-education institutions that
provide financial support and talent for modern technology, such as the AI industry.
They are efficient in advancing science and technology and providing new products,
while the wide application of AI products in those areas also leads to more AI-related
cases. The judicial practice of these cases provides experiences and references for
other regions. Through the in-depth analysis of these cases, we can find possible
solutions to the development of AI, reduce risks and promote the construction of
rules and systems in the field.

3.2 AI-Related Cases Are Less Common in Western

and Northeastern China

The number of AI-related cases in western and northeastern China is relatively small,
and some regions, such as Heilongjiang and Tibet, have zero such cases (this part
will only discuss the cases that have been disclosed given that some cases may not be
disclosed for the protection of trade secrets). The 2019 AI industry heatmap by the
Foresight Industry Research Institute shows that AI companies are mainly located in
the four metropolitan areas of the Beijing-Tianjin-Hebei region, the Yangtze River
Delta, the Yangtze River Delta, the Pearl River Delta, and the Sichuan-Chongqing
118 Y. Cui

Table 1 Top ten provinces

Locations Percentage Case Volume
with the most AI-related cases
Beijing 22.22% 126
Guangdong 18.17% 103
Hubei 10.41% 59
Jiangsu 10.23% 58
Shanghai 8.99% 51
Shandong 5.82% 33
Zhejiang 4.59% 26
Hunan 4.06% 23
Sichuan 3.35% 19
Anhui 1.59% 9

region. According to a report released by the Chinese Institute of New Generation

Al Development Strategies (CINGAI), as of 2019, the total number of AI enterprises
in China reached 745, with the Beijing-Tianjin-Hebei region, Yangtze River Delta,
Yangtze River Delta, Pearl River Delta, and Sichuan-Chongqing region accounting
for 44.8%, 28.7%, 16.9% and 2.60%, respectively.

4 Analysis of Litigants in AI-Related Cases

4.1 Diversified Litigants in AI-Related Cases

Most of the litigants in AI-related cases are natural persons (61.03%), legal entities
and other organizations (38.97%). Among them, legal persons and other organiza-
tions are mainly for-profit legal persons, but there are also cases involving special
legal persons, nonprofit legal persons and unincorporated organizations. The diver-
sity in litigants indicates the in-depth application of AI technology in all aspects of
society. In addition to noticing the different claims of enterprises and natural persons,
judicial professions also need to focus on protecting the legitimate rights and interests
of other types of subjects in judicial decisions.

4.2 There Have Been no Cases Where AI is a Litigant

AI technology at present is “weak AI”; it has the ability to apply intelligence to

one particular problem but lacks human consciousness. In judicial practice, “AI”
remains a tool. For example, in criminal cases, the subject of the crime is still a
person rather than a machine, and in the field of intellectual property, AI cannot
4 Al-Related Case Analysis (China) 119

autonomously generate inventions. In other words, so-called AI-generated inven-

tions rely on humans to operate the system. In addition, AI cannot independently
exercise the rights of inventors, such as the right of authorship and the right to receive
remuneration, and should prioritize personal decisions such as company policies.

5 Analysis of Litigation Types in AI-Related Cases

The growing number of AI-related cases brings challenges for the judiciary to identify
case types and unify the application of law in judicial practice.
As shown in Fig. 3, in civil, administrative, criminal, and law enforcements, AI-
related cases accounted for 72.43%, 9.79%, 4.24%, and 13.54%, respectively. Here
is a detailed analysis.

5.1 AI-Related Civil Cases Accounting for the Largest

Proportion

Most AI-related cases are civil cases, accounting for 72.43% of the total number.

5.1.1 Main Types of AI-Related Civil Cases

AI-related civil cases are mainly contractual disputes (35.7%), labor and personnel
disputes (23.8%), and intellectual property and competition disputes (23.3%).

5.1.2 Highlights.

Contractual disputes constitute most AI-related civil cases, and the contracts involved
were mainly technology contracts. For example, a telephone robot was involved in the
case of “Disputes over service contract between Shenzhen Hengfenghang Investment

Fig. 3 Types of AI-related

Cases
120 Y. Cui

Co., Ltd. and Shenzhen Shuxin Technology Co., Ltd.” (2018 Guangdong 0309 Civil
First Instance No. 6762 Civil Judgment, Longhua District People’s Court, Shenzhen
City, Guangdong Province). Technology companies that make light of legal risk
control will resort to legal disputes.
Labor and personnel disputes accounted for the second largest number of disputes,
mainly involving underpayment and withholding of promised labor compensation by
the employer, as well as disputes related to noncompetition. For example, in the labor
dispute between Microsoft China Co., Ltd. and Yu (Beijing Haidian District People’s
Court 2018 Beijing 0108 Civil Trial No.49790 Civil Judgment), a natural person Yu
who violates the noncompetition agreement, is faced with a high compensation of
more than 750,000 yuan in economic compensation for noncompetition and 1.2
million yuan in liquidated damages for noncompetition.
Intellectual property and competition disputes account for the third largest propor-
tion. Whether the content generated by AI is protected by copyright law has become
one of the hot issues in the industry. For example, in the case of “disputes over copy-
right, infringement and unfair competition between Shenzhen Tencent Computer
System Co., Ltd. In addition, Shanghai Yingxun Technology Co., Ltd.” (Shenzhen
Nanshan District People’s Court of Guangdong Province 2019 Guangdong Province
0305 Civil Preliminary Examination No. 14010), the court ordered that works gener-
ated by AI is in the scope of protection of copyright law, the work in the case was
completed by teams and labors of the plaintiff and the authorship of the work shall
belong to the plaintiff.

5.1.3 Advice

First, AI-related technology companies should pay attention to building a quality

legal team. AI has its drawbacks. It may bring risks while empowering society.
While taking the opportunity, technology companies should enhance their abilities
to prevent and handle legal risks relating to security, employment and innovation.
Second, AI-related employers should pay attention to human resource protection.
As a high-tech product, the commercial application of AI may generate risks. A
company may be at risk and even unable to pay for its employees if it expands
business blindly without a necessary feasibility study. Meanwhile, noncompetition
agreements, despite a common practice in the industry, are prone to disputes since
the field is full of fast-advancing and specialized technologies, and employees who
take legal risks lightly can lead to high compensation.
Third, the development of AI-related products needs to emphasize intellectual
property protection and fair competition. The data above indicate that with the number
of AI patents growing, infringements also increased. “AI will change the world”,
while embracing the new technology, it is essential for companies to stick to the
principle of honesty and business ethics, eliminate unfair competition and establish
a healthy market.
4 Al-Related Case Analysis (China) 121

5.2 Analysis of AI-Related Administrative Cases

Administrative cases accounted for 9.79% of AI-related cases.

5.2.1 Types of AI-Related Administrative Cases

Most administrative cases are administrative actions of China’s National Intellectual

Property Administration (NIPA) for rejecting applications of trademarks and patents.
NIPA could further clarify the approval criteria of trademarks and patents for a more
efficient AI trademark and patent registration.

5.2.2 Features

The trial of AI-related administrative cases calls for competent judicial professions.
For example, the administrative penalty over Wang, Xiamen Municipal Adminis-
tration for Market Supervision and Xiamen Yudian Automation Technology Co.,
Ltd. (this case was listed in the top 10 cases in intellectual property protection in
Fujian province by Fujian High People’s Court). The case involves infringements
upon trade secrets and administrative disputes between administrative subjects and
administrative counterparts, identification or evaluation experts and related experts
that appeared in the court. The focus of disputes was whether plaintiffs had infringed
on the third party’s trade secrets. Under the premise of being substantially identical
and contacting, the court analyzed the legality of the source in which the plaintiff
acquired technology related in the case and identified that the plaintiff’s claimed
buying “biochip technology” from another person was not an act of “reverse engi-
neering”. The court confirmed that in this case, the facts of a crime are clear and
evidence is hard and sufficient.

5.2.3 Advice

First, judicial research related to AI should be strengthened. Since AI-related admin-

istrative cases are highly specialized, people’s courts at different levels should empha-
size research and training, holding seminars on frontier theories and real situations
in the field to promote judicial capacity and efficiency.
Second, an integrated and sharing rule of law community should be established.
Expand exchanges and cooperation with government, colleges and universities and
practical organs and advocate the establishment of AI-related case trail systems.
122 Y. Cui

5.3 AI-Related Criminal Cases

Criminal cases account for 4.24% of AI-related cases.

5.3.1 Types of AI-Related Criminal Cases

AI-related criminal cases are mainly crimes where AI is a tool, crimes where AI is
a disguise and crimes invading personal information security.

5.3.2 Features

First, crimes using AI as a tool. For example, in the swindling case of Pan (Criminal
ruling No. 50, 04 Criminal Final Review, Liaoyuan Intermediate People’s Court, Jilin
Province, 2019), Defendant Pan represented the AI artificial intelligence calling
system to the defendant Zhang Group, recruited agents by using false platforms
such as “Yinghui Global”, “Golden Service”, “Ronghui” and “Yibao”, and settled
the defrauded funds through the fund settlement channels built by the false trading
platform.
Second, crimes using AI as a disguise. For example, in the case of pyramid selling
led by Zhao and Deng (Guangdong Province Dongguan intermediate people’s court
2019 Guangdong Province 19 criminal final judgment No. 757 criminal ruling), the
defendants refused to plead guilty, claiming that their innovation MoPai AI sharing
health robot was the world’s fifth invention and had made great contributions to
the country’s scientific and technological development as well as people’s well-
being. In Lian Meng Tian Xia, the pyramid selling company, there were 45 tiers of
pyramid schemes with a total of 76,459 people. Zhao was listed in the 12th tier, and
a total of 993 people from 15 lower tiers were headed by him. Combined with the
aforementioned witness testimonies, the statements of the codefendants and the trial
defendants, it is clear that Lian Meng Tian Xia is essentially a pyramid scheme in
the name of “pure capital operation”.
Third, crimes invading personal information security. AI products are highly
dependent on data. In the process of brutal data collection, it is easy for individ-
uals to violate the security of citizens’ personal information and even be directly
involved in dumping personal information. For example, in the case of infringement
of citizens’ personal information by Qiao and Wang (Jiangsu Changzhou interme-
diate people’s court 2019 Jiangsu Province 04 Final criminal trial No.50 criminal
ruling), Wang won success in both academic results and in his working field, he
has won several scholarships, honorary title of outstanding students and outstanding
student leaders, also holds patents and authorship in AI research, in spite of all the
achievements, he was jailed for invading personal information security.
As mentioned above, with the wide application of anonymization technology and
the internet, AI-related criminal cases are even more difficult to identify.
4 Al-Related Case Analysis (China) 123

5.3.3 Advice

First, more emphasis is placed on cracking down on criminal behaviors. To handle

AI-related criminal cases and support the implementation of a national AI strategy,
judicial organs should stick to strike criminal behaviors and pay special attention to
criminal investigation, criminal attacks and offender apprehension. Big data tech-
nology should be applied to analyze new types of criminal behaviors to keep on
attacking AI-related crimes.
Second, enhancing public legal education. Attach great importance to inform the
public on protecting their rights such as information security, and encourage all of
the society to respect, learn and obey the law, provide a favorable legal environment
for the development of AI.

6 Enlightenment from AI-Related Cases

As a key technology in the new round of technical and industrial revolution, AI plays
an important role in scientific competition. With more countries issued their national
AI strategies, AI technology in the global sphere has been constantly upgraded
and deeply integrated with other subjects, which profoundly reshaped political,
economic, social and national defense fields and has brought radical changes to
people’s lives and production.
In judicial practice, AI-related cases gradually show the characteristics of multiple
value judgments and a complex balance of interests. Given the above case study, the
further development of AI and rule of law should pay attention to the following
aspects.

6.1 More Emphasis on Analyzing the Features of AI-Related

Cases

The data analysis of AI-related cases may be conducted by big data analytics on
existing precedents, standards, and norms. Legislators may pay attention to the times,
locations, litigants and types of cases mentioned in this part, launch new standards
and norms through data collection and processing, and promote judicial interpretation
and litigation.
Governments have been working on AI legislation after noticing its implica-
tions for laws, ethical issues, society and the economy. In China, AI-related legisla-
tive items have been included in legislative programs (in a press conference held
at the second session of the 13th National People’s Congress in March 2019, the
spokesperson said that AI-related legislative items had been included in legisla-
tive programs). The Standing Committee of the National People’s Congress has
124 Y. Cui

included AI-related legislative items such as Digital Security Law, Personal Infor-
mation Protection Law and revision of law On Scientific and Technological Progress
in its five-year legislative program. Meanwhile, it also included AI legislation in the
urgent research project and conducted an in-depth investigation around relevant legal
issues to provide a sound legal environment for the development of AI.
We should further strengthen the analysis of the characteristics of AI-related
cases, identify new problems in a timely manner and keep up with the trend, provide
experiences and references for legislation, and contribute to building AI and rule of
law systems.

6.2 More Emphasis on Analyzing AI-Related Judicial

Decisions

6.2.1 Summarizing the Experience of Unifying AI Law Application

Promote unified law application by analyzing the application of certain legal provi-
sions through modern technology such as big data. Review experience on the appli-
cation of certain provisions in different judicial organs, combine the mainstream of
the application of laws and individual cases for a more accurate law application.
In addition, the data on the application of laws also shed light on legislation, and
legislative departments need to pay more attention to idle regulations and provisions
(Fig. 4).
The 5 most cited provisions are as follows:
(1) Article 60 of Contract Law of the People’s Republic of China: Each party shall
fully perform its own obligations as agreed upon.
The parties shall abide by the principle of good faith and perform obligations
of notification, assistance, confidentiality, etc. in accordance with the nature and
purpose of the contract and the transaction practice.

Fig. 4 shows the application of substantive law in AI-related cases

4 Al-Related Case Analysis (China) 125

(2) Article 107 of the Contract Law of the People’s Republic of China: Liabilities
for Breach of Contracts.
If a party fails to perform its obligations under a contract or its performance fails to
satisfy the terms of the contract, it shall bear the liabilities for breach of contract, such
as to continue to perform its obligations, to take remedial measures, or to compensate
for losses.
(3) Article 15 of Tort Law of the People’s Republic of China:: The methods of
assuming tort liabilities.
The methods of assuming tort liabilities shall include: cessation of infringement;
removal of obstruction; elimination of danger; return of property; restoration to the
original status; compensation for losses; apology; and elimination of consequences
and restoration of reputation. The above methods of assuming tort liabilities may be
adopted individually or jointly.
(4) Article 47 of the Labor Contract Law of the People’s Republic of China:
Calculation of economic compensation.
An employee shall be given economic compensation based on the number of years
he has worked for the employer and at the rate of one month’s wage for each full
year he worked. Any period of not less than six months but less than one year shall
be counted as one year. The economic compensation payable to an employee for any
period of less than six months shall be one-half of his monthly wages.
(5) Article 30 of Labor Contract Law of the People’s Republic of China: Labor
remuneration.
An employer shall, under the contractual stipulations and the provisions of the
state, pay its employees the full amount of remunerations in a timely manner. Where
the employing unit is in arrears or fails to pay labor remuneration in full, the laborer
may apply to the local people’s court for an order of payment according to law, and
the people’s court shall issue an order of payment according to law.

6.2.2 Improving Court’s Capacity for the Trial of AI-Related Cases

As a new type of case, AI-related cases require specialized and professional judicial
trials, and more lessons and experience should be acquired through practice.
AI-related cases have placed higher demands on the court’s professional trial
capability. Therefore, we should further improve the system of technical investigators
to address related complex cases. We should also strengthen the research and training
by offering courses on knowledge related to the frontier field of trial to improve the
competence of police officers and improve the quality and efficiency of the trial.

6.2.3 Issuing Model AI-Related Cases

Promoting unified law application is a highlight in comprehensive reform of the

judicial system and an important measure for the people’s courts to ensure judicial
impartiality and safeguard judicial authority. For the application of law in AI-related
126 Y. Cui

cases, judicial organs should pay attention to legal issues generated in practice and
actively address such problems. Promote the unification of judicial methods and
law enforcement standards through case guidance and the issuance of model cases.
According to case analysis, we identified several cases that shed light on practice.
For example, the “First Case of Facial Recognition” (Guo, an associate professor at
a Zhejiang university, sue Hangzhou Safari Park for Invasion of privacy) was ruled on
June 15, 2020. Facial recognition technology should be used under the supervision
of law and relevant departments, and authorities must introduce related laws and
regulations to safeguard individuals’ personal biological feature information.
The people’s court should strengthen case guidance, standardize case manage-
ment, place equal emphasis on handling cases well, summarize trial experience, and
adhere to the judicial accountability system and training trial experts. We should
improve the supporting mechanisms, such as the multiple discovery mechanism of
cases, the trial and guidance mechanism of cases, and the case quality manage-
ment mechanism, strengthen the trial business guidance, and promote the unified
application of the law.

6.2.4 Putting Judicial Big Data to Good Use

In an information society, judicial data represent judicial productivity. Good use of

judicial data will benefit trial procedures and the economy. By cooperating with
information technology companies, judicial organs could establish a data platform
to help with trial procedures and research, increase judicial openness, and improve
judicial credibility.
Judicial big data should focus on the establishment of law big data and case banks.
AI can learn from analyzed and systematized provisions and cases. Applying AI in
judicial practice can improve judicial quantity and efficiency.

6.3 Unifying Law Application in AI-Related Cases

Analyzing certain cases not only helps judges to make unbiased judgments and
portray economic development but also guides people’s behaviors and law enforce-
ment practice. For example, the analysis of civil cases can guide law enforcement
and supervision in industry and commerce; the analysis of administrative cases
can regulate the activities of state administrative organs in exercising administra-
tive authority, performing duties, implementing and enforcing laws in accordance
with legal authority and legal procedures; and the analysis of criminal cases can guide
the public security department on attacking criminal behaviors and apply targeted
measures.
4 Al-Related Case Analysis (China) 127

6.4 Establishing Diverse Mediation Systems for AI-Related

Cases

Dealing with AI-related cases calls for a balance between innovation and security,
freedom and justice since it involves different parties’ interests and values. Estab-
lishing a new form of mechanism for setting AI-related cases requires the efforts of
different organizations. Judicial organs need to form a comprehensive cooperation
system with government departments, industry associations, mediation organiza-
tions and tribunal organizations. For example, giving full play to the function of
entrusting professional institutions to mediate before filing a case and in litigation
and directly issuing enforceable mediation agreements through judicial confirmation.
AI technology should be applied in diversified dispute resolution systems, such as
“litigation and mediation connections”, to provide more efficient and highly qualified
mediation services.
Intensifying the entire mediation system is a major reform in advancing social
governance and promoting the rule of law at multiple levels and fields. In AI-related
diversified dispute resolution systems, lawsuits are not enough to meet the need for
right relief from market entities. We should promote the establishment of a civil
and commercial dispute resolution system by coordination between litigation and
nonlitigation to efficiently meet the needs of market entities and innovative social
governance mechanisms.
Chapter 5
Special Reports on the Development
of Artificial Intelligence and the Rule
of Law

Yadong Cui

1 Research on the Protection of Intellectual Property

Rights of Artificial Intelligence

1.1 Overview of AI Development and Intellectual Property

Protection

At present, AI has become the core driving force for a new round of sci-tech revo-
lution and industrial transformation, and it has an extremely profound impact on the
world economy, social progress and human life. As a strategic technology that steers
the course to the future, all major countries in the world have taken the development
of AI as a major strategy to enhance national competitiveness and maintain national
security, especially with an increasing emphasis on intellectual property and use it as
a tool to facilitate the development of AI. The development of AI and the protection
of intellectual property are intertwined. On the one hand, all countries attach great
importance to the research and development of AI technology, press ahead with the
application of AI-related patents and actively promote the development and applica-
tion of AI, striving to gain advantage in the new round of technological competition.
On the other hand, countries have introduced plans and policies on intellectual prop-
erty to further encourage the development of AI to gain the upper hand in future AI
competition.

Y. Cui (B)
Shanghai Law Society, Shanghai, China
e-mail: [email protected]

© Shanghai People’s Publishing House 2022 129

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2_5
130 Y. Cui

1.1.1 Taking Patents as an Example to Analyze the Development Trend

of Intellectual Property Protection of AI

In recent years, countries have attached great importance to the development and
application of AI technology, and AI-related patent applications in particular have
grown exponentially. According to the 2019 report of the UK Intellectual Prop-
erty Office (see Fig. 5.1), the number of AI-related patent applications has shown a
significant increase from 1998 to 2017.
In terms of application areas (see Fig. 5.2), the US has the largest number of
AI-related patent applications, followed closely by China.
Statistics from the China AI Patent Value and Competitiveness Report indicate that
as of November 2019, the number of AI-related patent applications in China reached
91,838. In view of the trend of such patent applications, the number of AI-related
patent applications in China was 562 before 2000; the number grew rapidly after
2011. In addition, as of November 2019, 25,431 AI-related patents have been granted,
with 81% of the grants going to Chinese applicants, and these AI-related patents cover
all provincial administrative regions in China. At the same time, Chinese Internet
companies have outstanding innovation ability in the field of AI and play a leading
role in the related technical fields.

Fig. 5.1 Number of patent applications

5 Special Reports on Development of Artificial Intelligence … 131

Fig. 5.2 Number of patent applications

1.1.2 Plans and Policies Introduced for Intellectual Property Protection

of AI by Countries and International Organizations

At present, countries and international organizations around the world have intro-
duced policies and initiatives related to AI in terms of investment planning, research
innovation, data privacy and others, which are mainly incarnated as follows:

China

China attaches great importance to the intellectual property protection of AI. In 2017,
the Development Plan for a New Generation of Artificial Intelligence by the State
Council proposed establishing a technical standard and intellectual property system
for AI, strengthening the protection of intellectual property in the field of AI, and
improving the mechanism to support the interaction among technological innovation,
patent protection and standardization in the field to promote the transformation of
AI innovations into intellectual property. Moreover, a patent pool for AI will be
established to promote the utilization and diffusion of new AI technologies. Reports
on the Work of the Government 2019 and 2018 both emphasized the need to accelerate
the development of emerging industries, promote the R&D and application of AI, and
cultivate a new generation of information technology and other emerging industrial
clusters to boost the digital economy.
In recent years, China’s judicial authorities have actively strengthened the protec-
tion of intellectual property in various fields, including AI. In April 2020, the Supreme
People’s Court released Intellectual Property Protection by Chinese Courts (2019)
132 Y. Cui

and the Opinions on Comprehensively Strengthening Intellectual Property Protec-

tion, aiming to improve intellectual property litigation procedures and trial mecha-
nisms, effectively curb intellectual property violations, and comprehensively enhance
intellectual property protection. In addition, on 1 January 2019, the Intellectual Prop-
erty Court of the Supreme People’s Court was inaugurated as a specialized body to
hear all technical IPR appeals nationwide. The Court not only issued the Key Judg-
ments of the Intellectual Property Court of the Supreme People’s Court (2019) in
April 2020, which conveyed the judicial philosophy, trial ideas and adjudication
methods for handling difficult, complicated and new types of cases in the field of
technical IPR but also propelled the “Zhiji” database of adjudication rules to be
operated online, promoting the unification of adjudication criteria for technical IPR
cases and improving the quality and efficiency of hearing technical cases. On 23
April 2020, the Shanghai High People’s Court released the Shanghai Courts’ Judi-
cial Protection Conditions for Intellectual Property 2019, proposing the working
goal to modernize, internationalize, professionalize and streamline intellectual prop-
erty justice to continuously increase and give full play to the effectiveness of judicial
protection of intellectual property, among others. All these initiatives and efforts by
China’s judicial authorities to strengthen the judicial protection of intellectual prop-
erty have created a good judicial environment and provided strong judicial support
and guarantees for the development of AI.

The United States

In 2011, the US introduced the National Robotics Initiative in an effort to establish

US leadership in next-generation robotics technologies and applications. 2017 saw
the introduction of a number of strategic decisions, including the National Robotics
Initiative 2.0 and the FUTURE of Artificial Intelligence Act, which was to further
advance the development and application of AI technologies in the US. In 2019,
the US signed the Executive Order on Artificial Intelligence Initiative to strengthen
national and economic security and maintain its R&D edge in AI and related fields.
In June 2019, the US National Science and Technology Council (NSTC) released the
report of the National Artificial Intelligence R&D Strategic Plan: 2019 UPDATE,
elaborating on long-term investments in AI, understanding and addressing AI-related
ethical issues, legal and social implications, and eight other sections on AI R&D
strategies.1

Japan

In 2015, Japan launched the New Robot Strategy, formulating a plan for the indus-
trialization of AI to speed up its R&D and application in key fields. In March

1National Science & Technology Council: The National Artificial Intelligence R&D Strategic Plan:
2019 UPDATE, June 2019.
5 Special Reports on Development of Artificial Intelligence … 133

2017, the Japan Revitalization Strategy and the Artificial Intelligence Technology
Strategy Council collaborated to develop an AI blueprint––Artificial Intelligence
Technology Strategy, setting goals for cooperation among the government, indus-
tries and academia in commercializing the achievements of AI technologies.2 In
2019, the Japan Integrated Innovation Strategy Promotion Council promulgated the
AI Strategy for All Individuals: Public, Industry and Government, pointing out that
the government should identify intellectual property issues in the process of AI
R&D and propose corresponding solutions in a timely manner. At the same time, the
Japanese government has actively released some examples of AI invention exami-
nation cases to further clarify the examination procedures on AI-related inventions
to promote effective examination of AI patents.3

Germany

In 2018, Germany introduced the Key Points for a Federal Government Strategy
on Artificial Intelligence, which proposed increasing funding for R&D and inno-
vation transformation in key areas related to AI and strengthening AI infrastruc-
ture. At the same time, the German federal government has developed an overall
policy framework for the development and application of AI strategies, including
strengthening the responsibility mechanisms of various departments and institutions
for the development of AI technologies, as well as introducing a number of poli-
cies and measures to actively encourage and support higher educational institutions,
public sector research institutes and small and medium-sized enterprises to apply for
AI-related patents to translate the technologies into applications where possible.4

The United Kingdom

In November 2017, the UK government published the Industrial Strategy in an effort

to put the UK at the forefront of the AI and data revolution. The UK government
has committed to a £995 million package to support the development of compa-
nies involved in AI technologies in the UK. In the meantime, effective measures
have been taken to promote and improve the transfer of intellectual property, and
the government has advised and facilitated universities and research institutions to
transform and apply and share their AI research achievements through intellectual
property. Moreover, the UK Intellectual Property Office has revised its examination

2 Yao Hanfei, Study on the Report of ‘Technology Trends 2019: Artificial Intelligence’ by WIPO,
Chinese Institute of Electronics, 2019(06).
3 Japan: Questionnaire on artificial intelligence police, at https://www.wipo.int/export/sites/www/

aboutip/en/artificial_intelligence/call_for_comments/submissions_march2020/ms_japan.pdf (Last
visited in Jun 19, 2020).
4 Germany: Questionnaire on artificial intelligence police, at https://www.wipo.int/export/sites/

www/about-ip/en/artificial_intelligence/call_for_comments/submissions_march2020/ms_japan.
pdf (Last visited in Jun 19, 2020).
134 Y. Cui

guidelines and related procedures on AI-related inventions to facilitate the effective

examination and granting of AI patents.5

The European Union

In 2018, the EU published the European AI Strategy, drawing up an EU plan of

action for AI to advance the EU’s position in the international AI competition. In
terms of the patentability of AI, the European Patent Office (EPO) issued the Guide-
lines on Patentability Criteria for Artificial Intelligence, which became an important
guideline on the patentability of AI technologies. On 28 January 2020, the Joint
Research Centre (JRC) of the European Commission published the report of Intel-
lectual Property and Artificial Intelligence, which studied and discussed the possible
implications and major controversial issues of the current development and adoption
of AI technologies in the field of intellectual property.6 On 19 February 2020, the
European Commission launched White Paper On Artificial Intelligence—A Euro-
pean approach to excellence and trust, which advocated establishing an ecosystem
of excellence and trust through talent acquisition, focusing on SMEs, promoting the
utilization of AI in the public and private sectors, and implementing the legal system
to advance the rapid development of AI in the EU.7

The WIPO

In September 2017, WIPO held initial consultations with a number of member states
on how WIPO should resolve AI issues. It took the lead in organizing the first Dialog
on IP and AI in September 2019, where they exchanged ideas on various topics
related to AI and the possible impact of AI on the IP system.8 Afterwards, the WIPO
drafted a list of issues on the intellectual property of AI, which provided the basis for
consensus on the main issues to be discussed or addressed in terms of IP policy and AI.
This list covers a wide range of substantive legal issues of IP policy arising from AI,
including patents, copyright and related property rights, data, designs, trademarks,

5 UK: Questionnaire on artificial intelligence police, at https://www.wipo.int/export/sites/www/

about-ip/en/artificial_intelligence/call_for_comments/submissions_march2020/ms_japan.pdf
(Last visited in Jun 19, 2020).
6 See IP and Artificial Intelligence: New Report by Joint Research Centre of the European

Commission, at http://www.iprhelpdesk.eu/news/ip-and-artificial-intelligence-new-report-joint-res
earch-center-europeancommission (Last visited in Jun 19, 2020).
7 See European Commission: White Paper On Artificial Intelligence—A European approach to

excellence and trust, Feb 19, 2020.

8 WIPO: WIPO Conversation on Intellectual Property (IP) and Artificial Intelligence

(AI), https://www.wipo.int/edocs/mdocs/globalinfra/en/wipo_ip_ai_ge_19/wipo_ip_ai_ge_19_2.
pdf (Last visited in Jun 19, 2020).
5 Special Reports on Development of Artificial Intelligence … 135

trade secrets, unfair competition and accountability for administrative decisions on

IP.9

1.2 A Theoretical Study of Intellectual Property Protection

of AI

At present, research on the intellectual property protection of AI has mainly focused

on AI technology and AI-generated products. However, AI technology is a technical
solution under patent law or software work under copyright law, and the existing
theories are sufficient to solve relevant problems. Therefore, most of the theoretical
studies focused on AI-generated products, mainly in the following two aspects:

1.2.1 Whether AI-Generated Products Are Protected by Existing Laws

In practice, the question of whether AI-generated products are technical solutions

under patent law or work under copyright law has become a hotly debated issue in
both theoretical and practical circles.

Affirmative Theory

The affirmative theory holds the view that AI-generated products constitute technical
solutions under patent law or work under copyright law. The theory is further divided
into objective standard theory and tool theory. Among them, scholars who support
“objective standard theory” believe that as long as an AI-generated product manifests
as a “work” from its appearance, it should be regarded as a work under copyright
law.
Obviously, to protect the works of AI-generated products, the “objective standard
theory” adjusts the traditional works theory in two aspects. First, “objective stan-
dard theory” expands the protected object of copyright law from human creativity to
machine creativity,10 and AI-generated products become works protected by copy-
right law. Second, “objective standard theory” objectifies the judgment standard of
existing works. As long as AI-generated products are the same as traditional works
in form, they can be protected.11

9 WIPO: Revised Issues Paper on Intellectual Property Policy and Artificial Intelligence, May 21,
2020.
10 See Dane E. Johnson, Statute of Anne-imals: Should Copyright Protect Sentient Nonhuman

Creators?, 15 ANIMAL L.15(2008), p. 27.

11 See Sun Shan, The Copyright Law Regulation of Artificial Intelligence Generated Content,

Zhejiang Academic Journal, 2018(02).

136 Y. Cui

The core of “tool theory” is to emphasize that the creators of works and the
inventors of patents should be natural persons, and AI can only play the role of
tools in the process of work creation and patent invention. Based on this theory, the
inventor must be human. Even if the AI system has a certain degree of autonomy, it
can only participate in the invention and creation as a tool and cannot become the
inventor in patent law. This view has been recognized to a certain extent in academic
circles.12 Compared with the simple application of “tool theory” in the patent field,
which directly interprets AI as the tool used by the inventor, some scholars of “tool
theory” not only emphasize that AI is the tool used by the creator but also emphasize
that the tool can reflect the will of human beings to a certain extent.

Negative Theory

Negative theory includes “nonlegal protection object theory” and “public domain
theory”. Among them, the typical one of “non legal protection object theory” is “non
work theory”.13 The theory holds that AI-generated products do not constitute works.
“Public domain theory” holds a similar view,14 but the difference between them is that
“non work theory” does not deny the necessity of legal protection of AI-generated
products, although it thinks that AI-generated products do not constitute works, while
“public domain theory” thinks that AI-generated products should belong to the public
domain without legal protection. Generally, the reason why the “negative theory”
thinks that AI-generated products do not constitute works is mainly the requirements
of the basic principles of copyright for originality. Originality requires that the work
must be the achievements of human intellectual labor, and the achievements should
reach a certain creative height. Thus far, AI-generated products are the results of
applying algorithms, rules and templates, which cannot reflect the unique personality
of the creator, so they cannot be recognized as works.15

12 See Lglesias, M., Shamuilia, S. Anderberg, A., Intellectual Property and Artifi-
cial Intelligence—A literature review, available at https://ec.europa.eu/jrc/en/publication/-
intellectual-property-and-artificialintelligence-literature-review, June 19 2020. WIPO,
Conversation on Intellectual Property and Artificial Intelligence, available at https:
//www.wipo.int/meetings/en/details.jsp?meeting_id=55309, June 19 2020. EPO, Patenting
Artificial Intelligence Conference Summary, available at https://www.epo.org/news-events/events/
conferences/2018/ai2018.html, June 19 2020.
13 See Wang Qian, On the Nature of the Content Generated by Artificial Intelligence in Copyright

Law, Legal Science (Journal of Northwest University of political science and law), 2017(05); Tao
Qian, On the Protection of Artificial Intelligence Achievements by Copyright Law, Law, 2018(04).
14 See Ralph D. Clifford, Intellectual Property in the Era of the Creative Computer Program: Will

the True Creator Please Stand Up?, 71 Tul. L. Rev. 1657(1997), pp. 1675–1680.
15 See Wang Qian, On the Nature of the Content Generated by Artificial Intelligence in Copyright

Law, Legal Science (Journal of Northwest University of political science and law), 2017(05).
5 Special Reports on Development of Artificial Intelligence … 137

1.2.2 How Should AI-Generated Products Be Protected

There are two typical forms of AI-generated products: the type of literature and art
and the type of technical solutions. At present, the relevant theories of AI do not
exclude the protection of these two types of AI-generated products. The difference
lies in the models adopted in protection.

The Protection Model of AI-Generated Products of Literature and Art

(1) Work Protection Model. Both “objective standard theory” and “tool theory”
hold that AI-generated products constitute works, and the difference lies in
the ownership of works under the protection model. According to “objective
standard theory”, AI should be regarded as a new type of civil subject, and
the copyright of AI-generated products should be enjoyed by AI.16 However,
most countries do not recognize the legal subject status of AI. Therefore, some
scholars suggest that AI should be modeled as a human being. As long as the AI
product constitutes a work in form, it should be recognized that AI has copyright.
Then, the judge could assign the copyright of AI to software developers, problem
solvers or computer owners.17 “Tool theory” holds that AI is only the tool used
by the creator, and the copyright of AI-generated products cannot be attributed
to AI, but there are differences in the specific attribution. Some scholars hold
that the copyright of AI-generated products should be owned by investors,18
some hold that the copyright of AI-generated products should be owned by AI
designers,19 and others hold that the rights of AI-generated products should
belong to users.20
(2) The Protection Model of Neighboring Right. Some scholars believe that
AI-generated products can be protected by neighboring right system. First,
although though AI-generated products do not belong to human creative intelli-
gence achievements, their forms of expression are consistent with the works
protected by the copyright law. Providing neighboring right protection for
them is conducive to the prosperity and development of literature, art and
science. Second, AI-generated products result from the substantial investment
of investors and have a certain market value. Therefore, providing neighboring

16 See Li Weimin, The Correct Nature of Artificial Intelligence Achievements in Copyright Law—a
Discussion with Professor Wang Qian, Oriental Law, 2018(03).
17 See Timothy L. Butler, Can a Computer be an Author? Copyright Aspects of Artificial

Intelligence, 4COMM./ENT. L. J. 707(1982), p. 744.

18 See Chen Mingtao, Wang Han, Research on Copyright of Artificial Intelligence Creation.

Copyright in China, 2017(03).

19 See Xiong Qi, Copyright Identification of Artificial Intelligence Generated Content, Intellectual

Property, 2017(03); Wang Wenliang, Wang Lianhe, An Examination of the Copyrightability of

Artificial Intelligence Creations in the Light of Law as Rhetoric, Technology and Law, 2017(02).
20 See Zhu Mengyun, Design of Copyright Ownership System of AI-generated products, Journal

of Shandong University (Philosophy and Social Science), 2019(01).

138 Y. Cui

right protection for them is in line with the development trend of intellectual
property law.21 According to this, a new kind of neighboring right can be estab-
lished for the works generated by AI by imitating the protection system of photo
neighboring right in copyright law in German.
(3) The Protection Model of Information Right or Special Right. Some scholars
believe that AI-generated contents can be the object of information rights protec-
tion,22 while some scholars hold that special rights can be created to provide
protection.23

The Protection Model of AI-Generated Products of Technical Solution

In terms of intellectual property protection of AI-generated products of technical

solution, the mainstream view is that AI cannot be the inventor. In January 2020, the
European Patent Office rejected the patent application of AI as the inventor.24 At the
same time, it will be a challenge for the patent law system to include AI-generated
products of technical solutions in the scope of protection under patent law. Generally,
the subject of creativity in patent law is ordinary technicians in the field, but the
knowledge and skills of ordinary technicians in the field should be improved when AI
is used as a tool.25 This improvement will affect the judgment of “nonobviousness”
because from the perspective of AI, many technological innovations will tend to
be obvious. Therefore, in the case of AI involved in invention and creation, if the
threshold of creativity for patents is not raised, patent proliferation will be inevitable.
However, an arbitrary increase in the threshold of creativity for patents would lead
to a decrease in patents for human inventions that do not rely on AI. There is no
consensus in the academic community on how to solve this dilemma.

1.3 Forward Thinking on IP Protection Solutions for AI

As a disruptive technology, the development of AI is bound to give rise to many

intellectual property issues. Based on some of the above arguments and scenarios,

21 See Wang Guo, On the Copyright Protection of “Works” Created by Computer, Journal of Yunnan
University, 2016(01).
22 See Guo Ruyuan, On the Protection of Information right of Artificial Intelligence Generated

Content, Intellectual Property, 2020 (02).

23 See Jani McCutcheon, Curing the Authorless Void: Protecting Computer-Generated Works

Following IceTv and Phone Directories, 37 Melb. U. L. Rev. 46 (2013), pp. 92—94.
24 See EPO, Patenting Artificial Intelligence Conference Summary, available at https://www.epo.

org/news-events/events/conferences/2018/ai2018.html, June 19, 2020.

25 See Lglesias, M., Shamuilia, S. Anderberg, A., Intellectual Property and Artificial Intelligence—

A literature review, available at https://ec.europa.eu/jrc/en/publication/intellectual-property-and-art

ificialintelligence-literature-review, June 19, 2020.
5 Special Reports on Development of Artificial Intelligence … 139

the author has made some reflections on the protection of intellectual property rights
of AI.

1.3.1 Reflections on the Scheme of Expanding the Subject of Rights

Under Copyright Law and Patent Law

This scheme advocates amending the Copyright Law and Patent Law, expanding the
creativity subjects of copyright and patent rights, to include the legal personality of
AI. On May 31, 2016, the Legal Affairs Committee of the European Commission
submitted a motion to the European Commission, which included positioning the
most advanced automatic robots as “electronic persons”.26 On 25 October 2017,
a “female” robot named Sophia was granted Saudi Arab nationality. However, the
biggest problem of that scheme is that it will bring unprecedented subversion to the
legislative value and institutional norms of Copyright Law and Patent Law. On the
one hand, the current Copyright Law and Patent Law protect the achievements of
human intelligence. Taking AI as the subject of copyright or patent rights will make
it difficult for copyright and patent rights to be self-consistent in theory since they
concentrate on stimulating human knowledge innovation. On the other hand, the
legislative models of both Copyright Law and Patent Law are based on the human
creativity subject. Expanding the subject of copyright and patent protection will not
only divide the original interests of human works and inventions but also damage
the system and regularity of the existing copyright and patent legislation, resulting
in considerable legislative costs.

1.3.2 Reflections on Setting up a Model for Protection of Neighboring

Rights of AI-Generated Products

The scheme advocates that AI-generated products should be protected within the
scope of neighboring rights. In civil law countries, the purpose of neighboring rights
is not only to protect the dissemination of works but also to protect investors who have
invested time, energy, technology, capital and other elements. For example, Germany,
Italy and other countries have set up neighboring rights for nonoriginal databases to
protect the interests of database investors. The neighboring rights protection model of
AI not only clarifies the boundary between human works and AI-generated products
in theory but also fits the basic function of the neighboring rights system to protect
the interests of disseminators and investors, which is conducive to the prosperity
and development of literature, art and science. It is one of the possible solutions

26European Parliament: Draft Report with Recommendations to the Commission on Civil Law
Rules on Robotics, May 31, 2016.
140 Y. Cui

to the legal protection of AI-generated products.27 However, there are also some
problems with this scheme. For example, the creators of performances, audio and
video recordings, broadcasting, nonoriginal photos, databases, etc., which are the
objects of traditional neighboring rights protection, are all human beings, and AI-
generated products are obviously incompatible with them. Moreover, the right subject
of traditional neighboring rights is relatively simple, while AI-generated products
involve investors, developers, users and multiple other stakeholders. In addition, if
AI is granted neighboring rights, how can we resolve the conflicts of interest between
AI-generated recordings, broadcasts and performances and traditional producers of
sound recordings, broadcast organizers and performers? These are the difficulties
and dilemmas that the neighboring rights protection model must face.

1.3.3 Reflections on the Scheme of Using Anti-unfair Competition Law

to Protect AI-Generated Products

The scheme advocates that AI-generated products should be defined as a kind of

investment interest, which should be protected by the nonright model of anti-unfair
competition law. In a series of cases, such as the International News Service case
and the National Basketball Association case, US courts have held that stealing other
people’s commercially valuable information may constitute unfair competition.28 AI-
generated products not only have commercial value but also may become competitive
advantages accumulated by investors through investment and labor. If others misap-
propriate AI products improperly, it may not only harm the original utilization of
investors but also affect the order of competition, which violates the basic principle
of honesty and credit. However, the anti-unfair competition law has its own system
and principles, which cannot regulate all the utilization of AI-generated products,
and its protection strength and scope are relatively limited, which also brings some
uncertainty. In addition, the protection model of the anti-unfair competition law
avoids basic issues such as the nature and ownership of AI-generated products, and
if we rashly apply the anti-unfair competition law to protect AI-generated products
without a clear understanding of the above problems, there is some doubt about its
effectiveness.

1.3.4 Reflections on Enacting a Special Law to Protect AI-Generated

Products

The scheme advocates enacting a special law to protect AI-generated products. The
advantage of the scheme is obvious. On the one hand, legislators can meticulously

27 See Xu Mingyue, Tan Ling, Discussions on the Protection of Neighboring Rights of Artifi-
cial Intelligence Creative crops-theoretical Proof and Institutional Arrangement, Comparative Law
Research, 2018(06).
28 See Li Mingde, American Copyright Law, Law Press 2014, pp. 675–690.
5 Special Reports on Development of Artificial Intelligence … 141

design the subject, content, ownership, limitation and liability rules of rights. On the
other hand, it can effectively balance the interests of various parties, including AI
investors, developers, users, the public and other subjects, to promote the healthy
development of AI technology and industry. However, the disadvantages of special
legislation are also obvious. On the one hand, special legislation on AI-generated
products would be costly. On the other hand, is the development of AI technology and
related physical industries in China mature enough? Is there sufficient understanding
and awareness of the application outputs of AI-generated products? Are the conflicts
among related interest groups prominent enough? These questions concerning the
legitimacy and necessity of separate legislation on AI have yet to be further examined
and certified.
In short, strengthening the intellectual property protection of AI has become the
consensus of all countries around the world, and it has also become an important
practice in many countries. At present, all kinds of arguments and suggestions around
the intellectual property protection of AI are based on clarifying the property rights
of AI and providing sufficient intellectual support for the healthy development of
AI technology and the physical industry. The key lies in finding the right point to
provide adequate protection for the development of AI while also promoting and
facilitating the development of intellectual property theory.

2 Research on Data Security in AI Application29

Currently, the development of AI is speeding up, which has prompted a flurry of AI

worldwide. Relying on the two basic platforms of big data and cloud computing, it
has established the three general technological systems of machine learning, pattern
recognition and human–computer interaction, gradually penetrating into the inno-
vation and development of major industries. In 2017, the global AI core industries
exceeded US$37 billion and are expected to exceed US$130 billion by 2020, with
an average annual growth rate of 60%.30 Moreover, its products and services are
widely present in all aspects of economic life, such as smart robots, intelligent
driving, security intelligence, smart homes, etc. At the same time, countries are
also attaching increasing importance to AI, actively engaging in forward planning
by introducing relevant policies and normative instruments. China promulgated the
“Development Planning for a New Generation of Artificial Intelligence” in 2017,
proposing that AI has become “a new focus of international competition” and “a
new engine of economic development”. The plan has been officially elevated to the

29 Author: Wu Shenkuo, Executive Director, the International Center for Network Rule of Law,
Beijing Normal University; and Secretary General, the Research Center of the Internet Society
of China.
30 Chinese Institute of Electronics: The New Generation Artificial Intelligence Development Plan

(2017).
142 Y. Cui

status of a national strategy, providing a solid backbone for the R&D, application
and promotion of this new generation of technology.
However, while AI holds great promise and a rich technological dividend, chal-
lenges and crises also arise. Many academics and experts have become increas-
ingly concerned about the security risks it poses: for example, at the 2017 Global
Mobile Internet Conference (GMIC), Stephen Hawking put forward The AI Threat,
saying that “It may become the terminator of human civilization unless we learn to
avoid the dangers”; Elon Musk, the Tesla CEO, has also warned that humans could
unwittingly “create an ‘immortal dictator’”. It has also been reported that AI can
increase existing threats as it can make attacks cheaper and faster, expand the range
of targets and assemble more targets; new threats will be introduced and AI systems
can be taken advantage of to perform forms of attack that humans cannot achieve; the
typical characteristics of a threat will be changed, making attacks more targeted and
more efficient and therefore impossible to response to the attacks and take practical
measures due to the difficulty of attributing the changes. The report also predicts that
AI will spawn new types of cybercrime, physical attacks and political subversion
in the next five to ten years.31 The above assertions illustrate the potential threats
posed by AI, not only in terms of physical attacks but also in terms of ethical, moral,
employment, political and other implications. It should be noted particularly that in
the field of cybersecurity, AI is also giving rise to a series of security risks against
system security, online content security, organizational security and management
security. The issue of data, which is closely related to online content security, has
become a key factor affecting the safe, reliable and controllable development of AI,
which needs to be given high priority and taken early countermeasures.
Before the specific analysis, there was a restriction to the scope of AI studied in
this paper. Internationally, it is common that AI could fall into three categories: weak,
strong and super, based on whether it can achieve human-like abilities in cognition,
thinking, reasoning, etc., or whether it has human-like free will or the ability to
think on its own. The three categories from weak AI to super AI can be considered
to be at the level of “human-like”, “human” and “beyond human”. However, we
are still in the early stages of weak AI. In the future, there is no dispute about the
possibility of strong AI and super AI being realized, only that the time of realization
is not yet determined.32 Therefore, without dwelling on general fantasies or being
overly concerned about unknown risks, this paper only takes weak AI into the study,
discusses the risks and criminal offences related to AI data at present, explores the
establishment of legislative norms in the context of the rule of law system of cyber
security, including criminal law and provides suggestions for the prevention and
control of risks in the industrial field.

31 See Future of Humanity Institute and others, “The Malicious Use of Artificial Intelligence:
Forecasting, Prevention, and Mitigation”.
32 See Calum Chace, [UK], SURVIVING AI, The Promise and Peril of Artificial Intelligence,

translated by Zhang Yaoran, China Machine Press, 2017, p. 118.

5 Special Reports on Development of Artificial Intelligence … 143

2.1 Risks and Crimes Related to AI Data

2.1.1 Infringements Related to Data Element

Data are the source of learning and decision making for AI. AI processes data of
large scale, variety, speed, timeliness and low value density through technical means
such as collection and storage, management and analysis, and visual computing. This
step is important, which makes the data element a major object of AI crimes. Data
security emphasizes its three qualities: confidentiality, integrity and availability of
data, which are also emphasized in AI applications. The following is an analysis of
the infringements to the data element, starting with the three qualities.
Infringements on data confidentiality. These infringements usually involve the
use of a user’s private personal information as a means of crime. For example,
criminals get to know about the assets status of an individual by stealing the user’s
bank statements and financial situation via Internet and willingness to pay based on
online behaviors, thus quickly and effectively targeting victims of financial fraud;
furthermore, criminals illegally obtain the victim’s personal private information,
composing it into multidimensional, mass, customized training data, and creating
malicious algorithms for automatically generating illegal websites, links, emails, etc.,
which are targeted and pushed to victims to entice them to click on, or committing
crimes in social network chats by learning this information disguised as the victims’
friends.
Infringements on data integrity. This prevents deep learning due to the lack of
critical data, often leading to the failure of the AI system to obtain information and
affecting the efficiency and accuracy of the next step. For example, a bot-driven,
large-scale attack named ‘information generation’ makes it extremely difficult to
obtain real information using tactics such as scattering information to disrupt the
normal operation of information channels.
Infringements on data availability. This generally includes data manipulation,
falsification and interference. For example, inputting nonrealistic training data to
an AI system to obtain results that run counter to the goal; altering, splicing and
creating fake videos or audios that are highly realistic to make inflammatory state-
ments through the effect of celebrities in order to disturb political and social stability;
creating malicious images in the process of image recognition by AI to trigger the
corresponding security bugs, change the control flow or data flow of the normal
execution of the program, and modify some key data such as tags and indexes so that
the AI system outputs the wrong results specified by the attacker.33
These three circumstances will become increasingly common in the future and
will have a significant negative impact on the AI systems themselves. In general,
the training of AI algorithms is based on a huge amount of data, and the system
collects more information with a wider range and more diverse content through
terminals, which makes the data more valuable in a material way, and where there

33See “40 Wise Insights on Internet Security (2017)”, edited by 360 Total Security, Posts & Telecom
Press, 2018, p. 296.
144 Y. Cui

is economic value, there is a tendency for negative effects arising from human acts
and consequently increasing security risks; moreover, AI systems are not indepen-
dent and exclusive. Rather, they are closely associated with the external environ-
ment and various elements. For example, in the process of researching and devel-
oping, producing, providing and applying a product or service, system designers,
platform providers, function managers and maintainers, as well as applicants and
other multiple subjects, are involved and have the appropriate qualifications to store,
modify, transmit, access and use the data according to their own identities, and this
flow of data between multiple subjects increases instability and ownership disputes.
In addition, the intelligence of AI can also bring external risks, as criminals can use its
technology to upgrade their criminal acts of violating personal information security,
which may result in the leakage of tens of thousands of citizens’ private information
and may also fuel the online black and gray market that has developed a clear top-to-
bottom relationship. Therefore, for many reasons, AI data-related infringements are
increasing, and those involved in the R&D, manufacture and provision of products or
services, and those who apply them, should pay greater attention to data protection
and give more efforts to security in all steps of storing, cleaning and integrating data
to prevent risks.

2.1.2 Typical Application Scenarios—Analysis of Data-Related Crimes

in Pattern Recognition

Pattern recognition is “the process of discriminating and interpreting patterns

(objects, behaviors, phenomena, etc.) contained in data by simulating the process of
human perception to analyze perceptual data such as images, speeches and videos”.34
Currently, it is widely used in biometric recognition, automatic speech recognition
(ASR), optical character recognition (OCR), remote sensing, medical diagnosis and
other fields. In the following, the author provides a specific analysis of data-related
crimes in pattern recognition, starting from several subapplication situations.
Biometric recognition has been developed for more than 50 years and is becoming
a more important approach to identify verification. For example, fingerprint recogni-
tion, which was first developed commercially, has the advantage of being convenient,
accurate and secure when verifying identity in switch-type systems; in the context
of the rapid development of deep machine learning in recent years, face recognition
has become a key tool in security monitoring by establishing massive amounts of
face image data and developing a series of efficient analysis models. In addition, as
a large number of smartphones and wearable devices enter people’s lives, biometric
recognition tends to develop portable applications and toward high-speed industrial-
ization. As such, biometric recognition is more dependent on both the user and the
environment, reflecting a deeper interaction between humans and the computer, and
its application mostly involves private data related to personal identity, which literally

34Chinese Association for Artificial Intelligence: White Paper on Pattern Recognition in China
(2015).
5 Special Reports on Development of Artificial Intelligence … 145

has higher requirements for security. In practice, biometric recognition is often asso-
ciated with financial and property crimes, with specific manifestations, including the
following: first, the problem of data leakage, where criminals can steal a large amount
of user privacy data, including fingerprints, irises, faces, body shapes and other iden-
tity information through attacking loopholes, which will be used to invade security
systems, steal money and goods from insurance equipment, and commit property
fraud; second, the problem of data tampering or destruction, where criminals can
use processing techniques such as deformation, blurring and true–false swapping to
falsify data and further interfere with the normal operation of data-based systems.
ASR has made great progress in the late 20th and early twenty-first centuries.
Supported by technologies such as model structure, deep neural network training,
multilanguage training and parametric learning, it has gradually achieved large-scale
industrialization and is playing an important role in voice services, voice evalua-
tion, security monitoring and translation. At the same time, technological enhance-
ments have also brought about security risks. In previous fraud cases, criminals often
confused their victims and gained their trust by impersonating their identities. Iden-
tity forgeries are mostly text messages, relatively simple and flat means, and many
people with high alertness do not easily fall into the trap. Currently, however, crim-
inals are learning ASR and developing voice synthesis systems to imitate the voice
of a specific subject to the extent that it is difficult to distinguish between the real
and the fake, and in the absence of other identity verification measures, it is easy to
lead victims into fraud.
Similar to ASR, visual identity systems (VIs) have a wide range of applications,
especially in security monitoring, such as the recognition of specific targets in the
crowd, vehicle recognition, object tracking, abnormal event detection, etc. VI relies
on image and video analysis technology, and the object of analysis is the massive
amount of data based on the Internet of Things, including all kinds of multimedia data
from surveillance cameras, mobile devices, the Internet, etc. By integrating different
data from different situations, it is possible for the system to analyze the behavior
with more complex targets, the correlation between targets and the evolution of
events between groups of targets.35 Data are therefore the key to VI, and its security
directly affects its effectiveness, accuracy and confidentiality. Recognizing this, an
attacker can compromise the functionality of VI by interfering with or corrupting the
data. A typical example is the implementation of an escape attack, where a specific
input sample is constructed to deceive the target system without changing the target
machine learning system.36 Such adversarial samples compromise the security of
online content to varying degrees, interfering with the operation of the VI and aiding
in the commission of further criminal acts.

35 See Chinese Association for Artificial Intelligence: White Paper on Pattern Recognition in China
(2015).
36 See “40 Wise Insights on Internet Security (2017)”, edited by 360 Total Security, Posts & Telecom

Press, 2018, p. 296.

146 Y. Cui

2.2 Legislation on AI and Data

2.2.1 Choice of Legislative Model

To determine the legislative model, that is, how to deal with China’s existing rele-
vant legislative norms and to what extent to enact new norms, some scholars have
proposed two dimensions for the cybercrime on choosing legislative model: one is
from a static perspective, considering the unitary model of penal code, or a plural-
istic model combined with a single-file only law related to cybercrime, or enacting a
comprehensive cyber law; the other is from a dynamic perspective, considering the
“progressive” or “forward-looking” way of legislation on cybercrime.37 The above-
mentioned perspectives can also be drawn upon for the construction of legislative
norms relating to AI and data. From the perspective of static dimension, the author
believes that it is more reasonable to continue with the unitary model of penal code
at this step. Although AI has been gradually applied in various situations and is
increasingly tied to various disciplines and industries, there is a real need to intro-
duce a blanket and systematic law on AI. At present, the US and the EU have made
attempts, with the US Congress proposing the FUTURE of Artificial Intelligence
Act and the European Parliament adopting a resolution on enacting civil law rules
on robotics. These legislative practices are still relatively preliminary and focus on
a small range of applications, with a view to further investigate and implement
supervision of AI. The current situation shows that the time is not yet ripe for a
comprehensive AI law, while the criminal law itself is very inclusive and evolving
and can respond to the needs of governance in the AI era by amending criminal law
and introducing laws and judicial interpretations. From a dynamic perspective, it is
recommended that a combination of “progressive” and “forward-looking” models be
adopted, whereby, on the one hand, the problem of AI risks is regarded as an objec-
tive phenomenon and legislation is mainly carried out by amending existing laws
and adding appropriate provisions, and on the other hand, the changes in criminal
acts, infringements of legal interests, scope of regulation and subject liability in the
future should be taken into account, leaving some room for judicial interpretation
and relevant laws and regulations.
It is undeniable that the development of AI has affected the existing legal relations
and brought about a series of new legal issues, which have challenged legislation,
enforcement and compliance. However, it would be too early to call for the imme-
diate enactment of AI law and the construction of a specific and specialized legal
system. Given the current state of AI technology, applications and the development
of industry, it is far from being out of control, and it would be inappropriate to call for
new legislation to deal with new crimes.38 For this reason, some of the security issues
that have arisen or may arise can be incorporated by combining and analyzing the

37 See Wang Yijue, Review and Reflection on China’s Cybercrime Regulation, Journal of Shihezi
University (Philosophy and Social Sciences), 2019(01).
38 Zhang Mingkai, The Criminal Legislation in the Internet Era, Science of Law (Journal of

Northwest University of Political Science and Law). 2017(03) Page: 69–82.

5 Special Reports on Development of Artificial Intelligence … 147

relevant existing legislative norms in China, amending existing provisions or adding

new judicial interpretations. Others can be regulated by conceiving and designing
future relevant legislative norms and formulating new legislative provisions from
both international and domestic governance perspectives.

2.2.2 Current System of Legislative Norms in China

The legislative norms related to data are defined in this article in a broad sense. First,
the protection of data is mainly concerned with data stored, used, processed and
transmitted in the system. This is not yet protected by our criminal legislation, which
is only reflected in Article 285(2) of the Criminal Law that the acts of unlawful access
to data are construed as crimes. With the emergence of more AI applications, human
facial data in the field of VI and vocal information in the field of ASR will become
criminal use, tampering and transmission, and other malicious acts of data processing
will also increase. It is necessary to take additional provisions into criminal law to
regulate illegal access, highlighting the importance attached to data protection. The
second is the protection of personal information, with more emphasis on the content
of privacy. It is important to note that the difference in term choice between countries,
where the term “information” is adopted in legislation concerning privacy in China,
while the term “privacy” is commonly used in the US and the term “data” has been
adopted by EU, which covers the meaning of privacy, since the introduction of the
Personal Data Protection Directive in 1995 with different terms corresponding to
different protection preferences and scope of protection. The following is a specific
analysis of the legislative norms of personal data protection.
With the advent of the big data era, people’s daily online activities generate huge
amounts of personal information, while at the same time, theft, misuse and malicious
disclosure are becoming increasingly serious. In response to this, relevant legislative
activities are being promoted in China. Not only has the Criminal Law established
the crime of infringing on citizens’ personal information, but the Cyber Security Law
has also set up cyber information security as a separate chapter; another example is
that the promulgation of the Decision on Strengthening the Protection of Cyber Infor-
mation in 2012 established the relevant principles of personal information protection
and regulated the collection, use and handling of citizens’ personal information by
websites and enterprises and institutions; the Regulation on the Protection of Personal
Information of Telecommunications and Internet Users introduced in 2013 improved
the protection of personal information of telecommunications and internet users. In
addition, in terms of local legislation, Guizhou Province promulgated the Regula-
tions on the Promotion of the Development and Application of Big Data in Guizhou
Province in 2016. As the first local regulation related to big data in China, it reflects
the government’s active action and high emphasis on the regulation of using data and
the protection of personal information.
It is clear from the above provisions that the protection of personal information
has become an extremely important part of legislation in China, and in an era where
AI is continuously developing, information security requires additional requirements
148 Y. Cui

and higher standards. To avoid the risk of data becoming a shortcoming that limits
the development of AI, the current laws for the protection of personal information,
especially the provisions of criminal legislation, need to adjust the parts that are
difficult to regulate according to the special nature of AI as soon as possible.
The protection of personal information in China’s criminal law dates back to the
enactment of the Amendment (VII) to the Criminal Law, which added one provision
of Article 253. It limits the subject to staff of state organs or important units, where
personal information originates from the performance of duties or the provision of
services by the units, with an emphasis on combating the sale, illegal provision and
acquisition of personal information. The amendment was a significant start in the
cause of legislative protection of personal information, but it did not elaborate on the
offence and did not clearly define the connotation of personal information, which
led to a number of disputes in judicial application. Moreover, due to the narrow
scope of the regulated subjects, it did not cover the common problem of information
leakage in the industrial field. The Amendment (IX) to the Criminal Law (hereinafter
referred to as “the Ninth Amendment”) has improved and formally established the
crime of infringement of citizens’ personal information. As the implementation of
the real-name system in reality has led to an explosion in the amount of personal
information and the sale of personal information has gradually formed a complete
criminal chain driven by profit, the Criminal Law has expanded the subject matter
from special to general to criminalize more acts that are in the upper reaches of the
crime. At the same time, the amendment clarifies the objective elements of the crime,
including taking any personal information illegally obtained from citizens into the
scope of the crime, which resolves the ambiguity of whether the “said information”
in the original provision refers only to information obtained from the work of special
units. In addition, the Ninth Amendment has increased the range of sentencing from
three to seven years and added the circumstances of heavier punishment, making the
setting of statutory penalties more reflective of the principle of suiting punishment
and responsibility to crime, so that a basic criminal law protection system has been
built with respect to citizens’ personal information, effectively responding to the
increasingly rampant crimes related to data in industry chains, and the acts involved
in AI crimes can also be basically incorporated into such a system. To further analyze
the adaptation of crimes against citizens’ personal information in the era of AI, the
following issues need to be highlighted:
(1) Does the concept of “personal information of citizens” need to be redefined?
Article 1 of the judicial interpretation published in 2017 clearly explains the
5 Special Reports on Development of Artificial Intelligence … 149

personal information of citizens.39 It covers information not only on identifica-

tion but also on the activities of a specific natural person, such as the where-
abouts, with special suggestive provisions, reflecting that the information is
exclusive, recognizable and valuable, that is, the information can directly or
indirectly identify a unique corresponding subject, and the information has the
legal interest protected by criminal law. This interpretation responds to the
practical needs of the fight against crime and helps the subjects of each unit
correctly determine the scope of duty of care of the subject who is compliant
with laws.40 In conjunction with the technical characteristics of AI, there is
no difference in the characteristics of citizens’ personal information under its
background, but some expansion in content may be taken into consideration;
for example, more record carriers or forms of recording may be enumerated,
and some forms of information related to AI may also be enumerated. In the
definition of personal information, the cyber security law mentions “personal
biometric information”.41 Similarly, with the promotion of biometric recogni-
tion, ASR, VI and other applications, people’s daily activities will generate more
diverse identity data, such as face images, fingerprints, and voice prints, and the
scope and frequency of using these data may even gradually surpass traditional
information, such as account passwords and ID numbers. Consideration could
therefore be given to reflecting this in the interpretation of the definition of
personal information.
(2) Will the multidirectionality, frequency, and complexity of the flow of personal
information under the support of artificial intelligence technology affect the
objective behavior of crimes and the setting of criminal law protection? Some
scholars have suggested that the current legislation mainly defines criminal acts
in terms of the “source” and “destination” of citizens’ personal information,
reflecting a protection path based on the horizontal flow of information but
less on the vertical use of information.42 As we can see from real-life cases,

39 Interpretation of the Supreme People’s Court and the Supreme People’s Procuratorate on Several
Issues concerning the Application of Law in the Handling of Criminal Cases of Infringing on
Citizens’ Personal Information, Article 1: ‘Citizens’ personal information’ as prescribed in Article
253A of the Criminal Law means all kinds of information recorded in electronic form or any other
form, which can be used, independently or in combination with other information, to identify a
specific natural person’s personal identity or reflect a specific natural person’s activities, including
the natural person’s name, identity certificate number, communication and contact information,
address, account password, property status, and whereabouts, among others.”.
40 Wu Shenkuo and Shi Jiali, “Criminal judicial protection of personal information in the context

of cybersecurity law”, Procuratorial Daily, 11 July 2017, p. 3.

41 Cybersecurity Law of the People’s Republic of China, Article 76(5): “personal information means

all kinds of information recorded electronically or by other means that can identify a natural person
individually or in combination with other information, including but not limited to the name, date
of birth, identity document number, personal biometric information, residential address, telephone
number, etc. of a natural person.”.
42 See Zhao Bingzhi, Zhan Qiwei, Reality Challenges and Future Prospects: A Reflection on Arti-

ficial Intelligence in Criminal Jurisprudence, Jinan Journal (Philosophy & Social Science Edition).
2019(01).
150 Y. Cui

AI facilitates personal information crimes and promotes the flow of personal

information, not only in one direction but also throughout the entire network,
which requires legislation to expand the path of protection to include new means
of infringement in the objective acts of crime.
(3) Are there new situations for the consent of victims when applying AI products
and services? The issue of victim’s consent in the criminal law protection of
personal information is in fact the issue of the subject’s right to informed consent
of personal information, that is, the subject’s right to be informed of the collec-
tion and use of his or her information, and the right to have his or her consent to
the processing of his or her information by anyone, unless an exception applies.
Article 3 of the Interpretation of the Supreme People’s Court and the Supreme
People’s Procuratorate on Several Issues Concerning the Application of Law
in Handling Criminal Cases of Infringing on Citizens’ Personal Information
clearly provides that the premise for the illegal provision of citizens’ personal
information is “without the consent of the person whose personal information
has been collected”. Article 41 and Article 42 of the Cyber Security Law also
emphasize the obligation to inform and consent to the collection, use and provi-
sion of personal information by network operators and further stipulate the right
of individuals to request deletion or correction in Article 43. In the context of AI
development, how can the right to informed consent of information subjects and
the freedom of data processing of operators in conducting business activities
be balanced? First, the right to informed consent currently covers mainly the
collection, use and provision of personal information, but with the increase in
the number of subjects involved in AI and the more diverse and complex ways
of accessing personal information, some new types of information processing
may emerge, or ways that were not entitled to be processed may be included in
the scope of processing with the individual’s consent due to the specificity of the
application. Second, in terms of the degree of protection of the right to informed
consent and the means of its realization, it is common for this right to be stipu-
lated in the Terms of Service and is mostly the prerequisite for registration and
use without giving users the right to choose, which may lead to the problem of
involuntary consent by the victim in judicial decisions, especially when increas-
ingly more smart products with technological features enter people’s lives. It is
necessary to clarify this point in the legislation as early as possible.

2.2.3 Prospects for Relevant Legislative Regulation

Before conceiving the path of criminal legislation, it is first necessary to determine

the principles and attitudes of the legislation. First, safety and development should
be taken as the guide to achieve balanced protection of the various legal interests
related to AI and innovation & development of the technology industry by crim-
inal law. Public power should adhere to the principle of technology neutrality when
intervening in the industrial field and should not overinterfere or suppress if the risks
5 Special Reports on Development of Artificial Intelligence … 151

incurred do not exceed social equivalence, control and tolerance.43 Second, criminal
law should be in accordance with the principle of necessity, maintaining rationality
in the midst of the novelty and frenzy brought about by technology, and strictly
maintaining its position as the “last line of defense”. At the same time, criminal law
in a society with risks should also be forward-looking, attach importance to poten-
tial risks and actively explore appropriate modes of early intervention, such as the
criminalization of preparatory acts for crimes. Third, attention should be given to the
continuity of the application of criminal law and the subsequent interface with other
criminal norms. As previously clarified, AI crimes will not fundamentally shake
the existing offence system of criminal law, and many acts can still be evaluated by
existing criminal law and judicial interpretations, either directly or with amendments,
so there is no need to blindly pursue a drastic legislative reform. In the process of
supplementary legislation, it is necessary not only to design criminal provisions that
are closely focused on the characteristics of AI crimes but also to consider the inter-
face with other criminal norms, especially the subsequent introduction of the law on
the prevention and control of cybercrimes and the relevant judicial interpretations.
Fourth, technical rules and legal rules are combined, reflecting the characteristics
of technical legislation. Currently, some general technical rules have appeared in
the process of application and development of AI, which should be referred to in
the legislation to determine the subject, object, means, amount and other content.
Moreover, AI is systematic and synergistic, and its crimes have the characteristics of
integrating objects, means and tools, and space, so its criminal acts and infringements
cannot be analyzed only from a one-sided perspective, and the legislative concept
should be carried out from the perspective of the whole crime and detailed connec-
tions as far as possible. With the guidance of these principles, the next step is to
envisage the specific legislative content, which can be carried out as follows:
From the practices of data protection in various countries, there are several ways
to legislation: the EU has enacted detailed acts such as the General Data Protec-
tion Regulation to protect data, which made a significant impact internationally; the
United States, mainly from the perspective of protecting the development of the data
industry, does not separately provide for personal data rights, but focuses on market
regulation and is mostly handled by the Federal Trade Commission in the event
of data infringement; and Germany has established a system of Federal Commis-
sioner for Data Protection, which specifically provides for the appointment of data
protection commissioners in relevant public institutions and individual organizations,
while China’s current legislation is mostly focused on regulating network informa-
tion services and information processing practices, emphasizing the protection of
personal information, while there are still many gaps in the regulation of specific
issues such as data rights.44 Drawing on the protection ideas of other countries and

43 See Zhang Quanyin, Opportunities and Challenges: Criminal Risks Brought by Artificial Intelli-
gence and Countermeasures of Criminal Law, Journal of Criminal Investigation Police University
of China. 2018(06).
44 Jiang Ye, Discipline of Algorithm and the Disciplined Algorithm: The Legal Regulation of the

Algorithm in the AI Era, Hebei Law Science. 2018(12).

152 Y. Cui

applying them to AI-related legislation, it is necessary to expedite the establishment

of a reasonable protection mechanism for personal information and data. With regard
to the protection of personal information, consideration should be given to redefining
its scope, distinguishing between sensitive information and other special types of
information and providing for separate forms of protection and restrictive content,
as well as further clarifying the principle of informed consent, etc., as explained
in the previous section and will not be repeated here. For the protection of data in
general, the current criminal law is “more static rather than dynamic”, the offences
mainly focus on the security of systems and personal information, and the only data-
related offence, “the offence of unlawful access to computer information and system
data”, is in fact also based on the protection of system functions and communication
security.45 The only data-related offence, “unlawful access to computer information
system data”, is also based on the protection of system functionality and communi-
cation security. In this regard, it is necessary to strengthen the importance of data
protection at the legislative level, such as clarifying the security management obli-
gations of relevant subjects, regulating their access, use and processing of data, and
establishing standard processes that are enforceable and can be used in different
applications; assessing whether it is necessary to add special rights such as the right
to be forgotten and the right to portability in terms of data rights; the EU Committee
on Legal Affairs recommends that, when formulating policies in the context of AI,
standards for the default protection of concepts such as privacy, informed consent and
encryption should be further developed.46 If necessary, specialized management or
research bodies can be established. For example, the UK government is working with
the Alan Turing Institute to establish a “Data Ethics Group” to study data science to
strengthen the scrutiny of data use. In addition, the law could extend its protection by
relying on AI to enhance protection. For example, it can try algorithmic agent models
that can set different permissions for data use depending on the circumstances and
manage both an individual’s consent and refusal to share information, which is both
for control and use at the same time, to facilitate communication among data-related
subjects.47

45 See Zhao Bingzhi, Zhan Qiwei, Reality Challenges and Future Prospects: A Reflection on Arti-
ficial Intelligence in Criminal Jurisprudence, Jinan Journal (Philosophy & Social Science Edition).
2019(01).
46 Cao Jianfeng, Top 10 recommendations! See how the EU predicts new trends in AI legislation

10, Robot Industry. 2017(02) Page: 16–20.

47 See China Electronics Standardization Institute, White Paper on Artificial Intelligence Standard-

ization (2018).
5 Special Reports on Development of Artificial Intelligence … 153

2.3 Strategies and Recommendations for Data Risk

Prevention and Control in the Industrial Sector

2.3.1 Basic Deployment to Carry Out the Safe and Scientific

Development of AI

In the industrial sector, the first task in carrying out the basic work of deploying the
safe and scientific development of AI is to establish a correct understanding of AI
and to adopt a dialectical and positive attitude toward this emerging technology. At
present, some people are overexaggerating the benefits of AI and are somewhat blind
to the application of the technology. Conversely, there are some people who have an
ignorant fear of technology and oppose the continued development of the industry for
fear that AI will eventually get out of hand. Both of these practices are undesirable. AI
is a “two-edged sword” that can drive economic and social development but can also
pose various types of security risks. Therefore, companies should first understand
the dual nature of AI and take into account two major orientations, development and
safety, when conducting business.
With development as the orientation, enterprises should have a detailed under-
standing of the external market, including the technology layout of similar enterprises
in the industry, the commercial application cases in each link of the value chain and
the predictable future development prospects, as well as assessing the application
opportunities based on their own business and the core competitiveness consisting of
their organization, technology and resources, and on this basis, they should formulate
a concrete and feasible development plan. In the process of execution, stimulating
internal innovation and leveraging external capital can accelerate the development
of AI.
With security as the orientation, companies should integrate security concepts
into all aspects of design, R&D, production and sales and adopt an approach of
“prevention beforehand, remediation afterwards” to avoid risks, with a series of
relevant principles and guidelines for action. For example, Microsoft has adopted
the six principles as the core of any deployment of AI-driven solutions, which are
privacy and security, transparency, fairness, reliability, inclusiveness and account-
ability.48 Another report proposes a scoring system for the foundation of AI devel-
opment in various industries, including first, organizational foundation, including AI
strategic vision and direction, talent and technical capabilities, organizational flexi-
bility and resilience; second, data, workflow and technology foundation, including
the amount of accessible data, maturity of data storage processes, data neatness, well-
documented and descriptive data, workflow automation, AI-friendly IT systems, etc.
Third, the implementation and application foundation includes the clarity of applica-
tion scenarios, the maturity of AI readiness, and the cooperation of solution service

48See Ralph Haupter, “AI is built on trust” https://news.microsoft.com/apac/2018/04/17/ai-is-built-

ontrust/ accessed 5 November 2018.
154 Y. Cui

providers.49 Some of the elements of the scoring system can also provide direction
for companies to guarantee AI security.
While referring to the above, particular attention should be given to accountability
and openness. Companies should establish a culture of responsibility by educating
their employees and emphasizing the importance of ethical statements and stan-
dards, the distinction between malicious use and use with good will of AI, and
awareness of risk prevention from the development of the technology to its applica-
tion. Researchers, in particular, have a responsibility to promote the beneficial use of
technology and prevent harmful uses, which can be practiced by working with policy
makers, providing expertise, and assessing the safety of projects. Openness, however,
is reflected in the appropriate disclosure of technical information, safety systems and
the sharing of experience and, where reasonably practicable, the promotion of public
participation and public decision-making. In addition, the industry needs to empha-
size the leading role of tycoons, which have a clear advantage in terms of development
and safety. These companies should take on more social responsibility and provide
a model for SMEs to follow.

2.3.2 Improving the Management and Control of AI Data

Ensuring the confidentiality, integrity and availability of AI-related data requires

an enterprise to have a clear data strategy, which may consist of the following
components:
Firstly, focus on the value of data assets. Building an ecosystem of data collec-
tion, integration and operation that serves the development of the business, taking a
programmatic approach to building data assets with the support of business segment,
while optimizing the collection, aggregation, utilization and subsequent updating of
data, and keeping it accurate, consistent and secure.50 In addition, from a develop-
ment perspective, ensuring the existence of open interfaces to digital systems is also
necessary to allow for future integration of technologies to flexibly accommodate
the development of AI.
Second, the protection of personal privacy should be strengthened. In the process
of collecting, using and processing data, the relevant personnel need to adopt appro-
priate technical means to prevent the leakage, tampering and destruction of personal
privacy information, such as data encryption, by means of “homomorphic encryp-
tion” to effectively enhance the security of information transmission51 ; data distortion

49 See Chinese Association for Artificial Intelligence, Roland Berger, White Paper on Innovative
Applications of Artificial Intelligence in China (2017).
50 See Chinese Association for Artificial Intelligence, Roland Berger, White Paper on Innovative

Applications of Artificial Intelligence in China (2017).

51 Homomorphic encryption allows a specific algebraic operation to be performed on the ciphertext

and the result is still encrypted, in the same way that the same operation is performed on the plaintext
and the result is encrypted. See 360 Total Security, “40 Wise Insights on Internet Security (2017)”,
Posts & Telecom Press, 2018, p. 110.
5 Special Reports on Development of Artificial Intelligence … 155

can also be carried out, such data can still maintain some properties and makes it diffi-
cult for an attacker to recover the true original data from it; restricted release can also
be used to selectively release sensitive data and related information by balancing
the risk of disclosure with ensuring accuracy through “data anonymization” tech-
niques, which may also involve a hierarchy of importance for the data. In addition,
an authentication management system can be established, including authentication
of user identity, application identity, device identity, etc., to raise the threshold for
privacy data leakage.
Third, a data self-protection system can be set up to monitor the neatness and
security of the data in real time through programming so that it is not disturbed by
the external environment, ensuring that the core data are capable of resisting all forms
of malicious attacks at any time and in any place.
With regard to the protection of data security, special attention should also be paid
to the requirements of the General Data Protection Regulation (GDPR). A study has
identified nine major impacts of the GDPR on AI: Article 6 prohibits the reuse of data,
which limits the innovation of AI development and use to a certain extent; Article
17 provides for the right to be forgotten of data, which is contrary to the requirement
of vast amounts of data by AI machine learning, which may affect the accuracy of
AI and even cause some damage. Article 20 provides for the right to portability of
data, which could stimulate competition among companies by promoting the sharing
of relevant data between consumers and AI companies, and the localization of data
required by Articles 44 to 50 could significantly increase the cost of processing data
by AI. In addition, the lack of clarity on data identification standards under the GDPR
will also discourage companies from using unidentified data and narrow the scope of
its possible legitimate uses.52 As such, the GDPR will have a significant impact on
the AI industry in data processing, and all relevant companies under the supervision
of the GDPR should take early steps to improve their business compliance to adapt
to these impacts.

2.4 Conclusion

Just as the criminal law made a positive response to the increasing cyber crime, the
criminal law also needs to control the new criminal risks brought by artificial intel-
ligence at the right time and in the right way. In order to deal with the limitations
in criminal culpability and penalty system, and the difficulty in identifying harmful
behavior and causality, the criminal law needs to be adjusted and transformed in time
to realize its value in the new era.53 Obviously, the risk management of AI data is

52 See Information Technology and Innovation Foundation, “The Impact of the EU’s New Data
Protection Regulation on AI”.
53 See Liu Zhiwei, Achievements and Prospects of China’s Criminal Law Research over the Past

Forty-year Reform and Opening up, Journal of Hunan University of Science & Technology (Social
Science Edition). 2018(05).
156 Y. Cui

complex, and broad social in nature, and thus cannot be solved by legal means alone.
We also need to break down disciplinary barriers in academia, strengthen technical
communication in industry, and actively engage in dialogue between the govern-
ment and the public, so as to include the government, enterprises and institutions,
experts, technicians and the public in a pluralistic and integrated AI security gover-
nance mechanism, and establish a pluralistic and interactive risk management system
including legal regulation, ethical regulation, industry regulation and self-regulation,
so as to create a healthy ecological environment necessary for the development of
AI innovation.54

3 Research on Legal Issues of Facial Recognition

and Privacy Protection55

Currently, facial recognition technology has been used in various areas and has shown
great potential. In 2016, China’s facial recognition industry market exceeded 1 billion
yuan. In 2019, the global market for the industry was $2.391 billion and is expected
to reach approximately $5.1 billion in 2021. Among its development, application
and expansion, the technology also poses challenges for law, security and ethics,
particularly in personal privacy protection, which has attracted much attention.

3.1 Facial Recognition Technology and Its Application

Scenarios

Facial recognition technology refers to a technology that captures images or video

streams containing human faces through a tracking camera, automatically detects
and tracks human faces, and then identifies them. The technology usually includes
three parts: face detection, face tracking and face comparison.56 With the rapid devel-
opment of facial recognition technology applications, it has penetrated various areas,
especially in the fields of finance and security.57

54 See Ma Changshan, Social Risk and Legal Regulation of Artificial Intelligence, Science of Law
(Journal of Northwest University of Political Science and Law). 2018(06).
55 Author: Cui Yadong, Secretary of the Party Group and President of Shanghai Law Society, (Grand)

Justice of Second Rank; Yang Hua, Professor and Doctoral Supervisor, Dean of School of Artificial
Intelligence and Law, Shanghai University of Political Science and Law; Hu Rongxin, Cadre,
Shanghai High People’s Court Office; Zhang Sundong, Master’s student, Shanghai University
of Political Science and Law.
56 See Li Huaijin: Legal Defense of Facial Recognition Technology Risks. Retrieved from http://

news.cssn.cn/zx/bwyc/201903/t20190306_4843411.shtml?COLLCC=4006574999&.
57 See Liang, Li Wen, Facial Recognition Privacy Protection Standards Need Urgent Improvement,

Fintech Times, No. 11, 2017.

5 Special Reports on Development of Artificial Intelligence … 157

3.1.1 Overview of Facial Recognition Technology

Currently, facial recognition technology mainly refers to the technical means that
recognize a “human face”, which is essentially the extraction and use of personal
biological information. The main features of facial recognition technology are as
follows. First, contactless. It may capture face information without any active coop-
eration from the person. An ordinary camera serves as the sensor for face information
collection, and facial recognition software and algorithms analyze the image so that
long-distance information can be captured and stored.58 Second, strong autonomy.
It works smoothly even when individuals show no signal. Third, intuitive and effi-
cient. It simultaneously sorts, recognizes and analyzes faces of multiple individuals
in the same scene, which greatly improves the recognition efficiency.59 Fourth, open-
ness and asymmetry of the technology application. The capture of face biological
information can be achieved at a very low cost by different individuals, there is a
strong information asymmetry between the collector and the collected party, and the
process can be performed without any notification of the collected party; therefore,
the collected party may not be aware that his face pattern was scanned and collected.60
The technology has a relatively low threshold for both computer software and hard-
ware, and the process of capturing face information is convenient. It will certainly
have more application scenarios in the future.

3.1.2 Major Facial Recognition Technology Application Scenarios

Application in Public Security

In the field of public security, from the “Skynet” project to the Xue Liang campaign,
mass video surveillance programs have been launched by the public security system.
The Internet use of public video surveillance systems integrates various videos
and images and will contribute to urban and rural social governance, intelligent
transportation, people’s livelihood, and ecological construction and protection.61

58 See Yang Jian: Legal regulation of Facial Recognition technology application. Retrieved from
https://www.cssn.cn/zx/bwyc/202003/t20200325_5105433.shtml?COLLCC=2158995962&.
59 See Agricultural Bank of China, Nanjing Jianye Sub-Branch Group, Research on Privacy Protec-

tion and Security Application of Facial Recognition in Commercial Banks, Modern Finance, Vol.
12, 2019.
60 See Yang Jian, Legal regulation of Facial Recognition technology applications, China Social

Science Journal, March 25, 2020.

61 See Zhou Mingyao, Facial Recognition Application of “Xue Liang” Project”, Intelligent Building

and Smart City, 12, 2019.

158 Y. Cui

Application in Social Life

In the field of social life, from unlocking smartphones, checking-in for trains, auto-
mobiles and airplanes, to paying with face in shopping malls, facial recognition
technology is used everywhere.62

3.1.3 The Legal Nature of Facial Recognition Technology Application

From the viewpoint of facial recognition objects, technology by its nature refers
to the collection, storage and analysis of personal biological information through
algorithms and computer programs. Therefore, the legal nature of such technology
shows the following features. First, facial recognition information is personal iden-
tity information that can be “directly identified”. From the perspective of legal rights,
information captured through facial recognition technology belongs to civil rights
and should be included in privacy law. Second, legal regulations of facial recognition
technology vary in different fields. For example, in the public security field, certain
authorities shall have the right to directly obtain personal information, including face
information for social security, criminal investigation and national security; in the
commercial field, the attainment of face information must be based on individual
privacy rights, and any sharing, alienation and deposition of such information shall
be restricted without obtaining prior authorization from the customer.63 The legal
conflicts of facial recognition technology are mainly reflected in the fact that norma-
tive legal standards cannot fully guarantee the legal rights of individuals. At present,
China has not yet introduced a complete legal supervision system for facial recog-
nition, and issues such as technology abuse and insufficient protection of personal
privacy have occurred in the commercial application of facial recognition.

3.2 Protection of Privacy in Facial Recognition and Its

Related Cases

The right to privacy belongs to the category of civil rights and is a specific personality
right, which has connotations such as physiological information, physical health,
genes and so on.64 As a new technology, facial recognition uses face information
as biological information, therefore giving rise to legal controversies in privacy
protection.

62 Forward-looking Industry Research Institute: China’s Facial Recognition industry market status
and development trend analysis in 2020 three-dimensional Facial Recognition will prevail. Retrieved
from https://bg.qianzhan.com/trends/detail/506/200211-163d0500.html.
63 See Zhu Wei, The Legal Nature of Facial Recognition Determination, Procuratorial Daily,

November 6, 2019.
64 See Wang Liming, Redefining the Concept of Privacy, The Jurist, 1, 2012.
5 Special Reports on Development of Artificial Intelligence … 159

3.2.1 Privacy Protection in Facial Recognition Application

Privacy means a lot for individual citizens, and Article 29 of Law of the People’s
Republic of China on Protection of Consumer Rights and Interests stipulates that state
organs concerned shall, in accordance with the provisions of laws and regulations,
punish any law-breaking or criminal activities of business operators infringing upon
the legitimate rights and interests of consumers in their supplying commodities or
services. At present, the main focus of privacy protection in the facial recognition
field is as follows:

Public Rights and Privacy Protection of Facial Recognition

With the advent of the era of AI, the invasion of personal privacy has shifted from
the traditional “tortfeasor” to various electronic terminals. The infringement itself
will also be conducted in a more intelligent and secrete way. Information can be
collected without the person’s knowledge, especially in the public space.65 From the
perspective of the public space, personal privacy should be protected even in the
public environment because there is a difference between being seen by others in
public and being photographed by an unknown observer and having the image saved.
A sneak shot may include more detailed features, which are prone to secondary
processing and constitute an infringement of privacy. Generally, the problems of
public rights and personal privacy caused by facial recognition technology are not
completely incompatible, and win–win results can be achieved through legal means.

Legality of Face Information Collection, Storage and Use

At present, according to their major application field, the collection, storage and use of
face information are roughly divided into two categories: for public use and for private
use.66 More specifically: First, for social security management. For example, the
public security system has launched the “Skynet” project, a nationwide surveillance
system for tracking criminals, which helps collect and store face information and
helps to combat crimes. Second, for enterprises, face information is used to improve
competitiveness, brand attractiveness and customer experience. For example, facial
recognition payments are widely used across China. Third, for research institutions,
the technology can be used in academic resource collection. Fourth, for social orga-
nizations, they collect and store face information for management and operation. For
the reasons above, concerns and controversies arise regarding whether collecting face
information by facial recognition technology is legitimate, mainly in the following

65 See Xu Tianying, Privacy Dilemma and Relief Path in the Age of Artificial Intelligence, Journal
of Southwest University for Nationalities (Humanities and Social Sciences Edition), No. 6, 2018.
66 See Jie Jiang, Infringement Risks and Control Strategies of Facial Recognition Technology

Applications, Books & Intelligence, Vol. 5, 2019.

160 Y. Cui

areas. First, for needs of public interest (mainly for public safety), authorities may
conduct information processing in accordance with the law. This situation generates
less disputes.
Second, face information is a tool for private purposes, which in practice holds
more disputes on legality. At present, China has not issued specific legal documents
for facial recognition information and its technical application, and therefore, prob-
lems can only be solved by referring to relevant laws. For example, article 29 of
the Law of the People’s Republic of China on the Protection of Consumer Rights
and Interests stipulates that “In collecting and using the personal information of
consumers, business operators shall adhere to the principles of legality, rationality
and necessity, explicitly state the purposes, methods and scope of collection or use
of information, and obtain the consent of consumers.”

3.2.2 Domestic and Overseas Facial Recognition-Related Privacy

Protection Cases

The Disputes Over Guo Bing and Hangzhou Wildlife World for the Improper
Use of Facial Recognition Technology

Guo Bing, a professor at Zhejiang Sci-Tech University, purchased the Hangzhou

Wildlife World Annual Card for 1370 yuan, which allowed him to enter the zoo by
fingerprint recognition. In July 2019, the Wildlife World changed the way for annual
card customers to enter the zoo from fingerprint recognition to facial recognition.
Guo believed that the zoo had violated their privacy and refused to apply for facial
recognition. Then, Guo was rejected by the wildlife world, and he filed a lawsuit to
the court against the zoo. The case was called the “First Case of Facial Recognition”
in China. From the perspective of commercial subjects, although using facial recog-
nition technology for admission is not prohibited by law, it is difficult to justify park
behavior. In addition, the park lacks the capacity to ensure the security of customer
face information, which ultimately has led to the lawsuit.

Tang and Zhang’s Larceny Case of Using Facial Recognition Technology

Improper use of facial recognition technology brings about civil disputes, and its
improper and illegal use may even bring criminal risks. In 2019, the People’s Court
of Chengdu Pidu District heard a facial recognition-related larceny case. In this
case, Tang obtained the victim’s Alipay account and face information through illegal
means, passed the Alipay facial recognition program by a 3D dynamic digital image,
and provided the account to another suspect, Zhang, who then committed the crime,
stolen a total of more than RMB 24,000 from the victim’s Alipay account.67

67See 2019 Sichuan 0124 Criminal First Instance No. 610 Criminal Judgment, Pidu District People’s
Court, Chengdu City, Sichuan Province.
5 Special Reports on Development of Artificial Intelligence … 161

The Class-Action Lawsuit of Google Users V. Google Inc. For Information

Leaking

In December 2018, the plaintiffs claimed that Google’s facial recognition feature
with photosharing and storage infringed on their privacy and proposed a settlement
of more than $5 million, which was ultimately rejected by the court. The court held
that plaintiffs have not suffered an injury sufficient to establish Article III standing
and that their claims are dismissed.68

3.3 The Legislation of Privacy Protection in Facial

Recognition

Many countries have attached great importance to privacy protection in facial recog-
nition and have implemented relevant laws and regulations, especially in fields where
facial recognition technology is applied as well as information collection and storage.

3.3.1 The Legislation of Privacy Protection in Facial Recognition in EU

The European ePrivacy Directive 2002 provides that users must be informed before-
hand of information collection and have the option to opt out. However, in its 2009
revision, the provision was amended to “to allow the use of information on the user’s
terminal only if the user is provided with the explicit purpose of its use and complete
information, and if prior permission is obtained from the user”. In January 2012,
the European Union issued a draft amendment to the “Data Protection Directive”,
in which it was proposed that no processing of user data is allowed without the
user’s consent. A data protection officer should be designated in an enterprise that
employs 250 persons or more; otherwise, the enterprise may face severe punish-
ment.69 In 2018, the EU’s General Data Protection Act (GDPR) came into force, and
in November 2019, the EU Fundamental Rights Agency released Facial Recognition
Technology: Fundamental Rights Considerations in the context of Law Enforcement
and called for a clear regulatory framework of facial recognition technology.70

68 See Wang Danna, Biometrics: Innovative Application of Traditional Information Security in New
Technology Environment, China Information Security, No. 3, 2019.
69 See Duan Weiwen, Ji Changlin, The Right to Privacy in the Era of the Internet and Big Data,

Science and a Society, Vol. 2, 2014.

70 See Wang Junxiu, Privacy Remodeling in Digital Society: The Case of “Facial Recognition”,

Exploration and Controversy, No. 2, 2020.

162 Y. Cui

3.3.2 The Legislation of Privacy Protection in Facial Recognition

in the United States

In the United States, there is no federal law on facial recognition; instead, each
state has separate regulations on biometric data. For example, in August 2019,
Massachusetts lawmakers voted to banning the government and police use of facial
recognition and stipulated that evidence obtained from facial recognition technology
was inadmissible in litigation in the state. In contrast, the Illinois Biometric Infor-
mation Privacy Act, also the first bill to regulate biometrics in the United States,
provides regulations on the use of three types of biometric information: first, in
the initial collection of personal biological information, the information subject
should be informed of the content, utilization, purpose and retention time of the
collected information; second, the private entities “in possession” of biometric iden-
tifiers/information must develop a retention policy; and third, private entities in
possession of biometric identifiers/information may not “disclose, redisclose, or
otherwise disseminate” the biometric information unless they obtain consent from
the individual or his/her legally authorized representative.71
The U.S. Congress has not yet established any federal law on the issue. The Algo-
rithm Accountability Act of 2019 stipulates that the development and application of
computer algorithms of all types shall be regulated and held accountable if any defect
exists, and companies are explicitly required to take responsibility for fixing defective
algorithms to avoid biased, discriminatory, or inaccurate output.72 The Commercial
Facial Recognition Privacy Act of 2019 was included in the Senate Committee on
Commerce, Science, & Transportation as a model provision for regulating facial
recognition technology, products, and services in the United States.73 In addition,
cities such as San Francisco and Oakland have banned government face surveillance
in public places; California stipulates that facial recognition for commercial purposes
shall be equipped with obvious and specific markers; Massachusetts, Berkeley and
New York City have joined the discussion of banning facial recognition and other
remote biometric surveillance.

3.3.3 The Legislation of Privacy Protection in Facial Recognition

in China

In recent years, new technologies such as the Internet, big data, artificial intelli-
gence, and blockchain have powered social and economic progress. At the same time,
advanced technology, the “double-edged sword”, also brings risks and challenges.

71 See Zhou Kunlin, Li Yue, Research on The Legal Regulation of Face Data Use Under Response-
Based Theory, Southwest Finance, No. 12, 2019.
72 See Algorithmic Accountability Act of 2019[DB/OL]. [2020–04-10]. Retrieved from https://

www.congress.gov/bill/116th-congress/house-bill/2231.
73 See Commercial Facial Recognition Privacy Act of 2019. Retrieved from https://www.congress.

gov/bill/116th-congress/senate-bill/847, last visited 10 April, 2020.

5 Special Reports on Development of Artificial Intelligence … 163

With the rapid development of new technologies, China attaches more importance
to regulating and protecting technological advances through rule of law thought and
methods, which is also true for facial recognition technology.
For example, in December 2017, the National Information Security Standard-
ization Technical Committee issued the National Standard on Personal Information
Security (implemented in May 2018 and revised in 2019 and 2020). The document has
included biometric information as personal sensitive information. Explicit consent is
required in the collection of such information, the minimization principle is stipulated
in information processing and transformation, and the new PI controller shall obtain
another explicit consent from the PI subject; specific disposal procedures are also
stipulated for security incidents.74 In December 2015, the National Standardization
Administration of China and the General Administration of Quality Supervision,
Inspection and Quarantine issued The Technical Requirements for Security Video
Surveillance Facial Recognition Systems (GB/T 31,488–2015). On November 7, the
24th meeting of the Standing Committee of the 12th National People’s Congress
passed the Cybersecurity Law of the People’s Republic of China. The acts mentioned
above provide specific regulations on the collection and use of personal informa-
tion for online operators. Regulations on Facial Recognition also appeared in the
Public Security - Facial Recognition Application - Image Technology Requirements
(GB/T 35,678–2017), which is included in the 32nd Chinese National Standard
Notice, published by the General Administration of Quality Supervision, National
Standardization Administration in December 2017.
The Civil Code of the People’s Republic of China, adopted at the Third Session
of the 13th National People’s Congress on May 28, 2020, also stipulates in Article
1019 that “No organization or individual may infringe upon any other person’s right
to portrait by vilifying, defacing, forging by means of information technology or
otherwise. Without the consent of a person entitled to portrait, the portrait of such
a person shall not be produced, used or made public unless otherwise provided by
law. Without the consent of a person entitled to portrait, the holder of rights to
portrait works shall not use or make public the portrait of such person by means of
publication, reproduction, distribution, lease, exhibition, etc.”, which fully conveyed
China’s determination to protect facial recognition privacy rights from a legal way.
Globally, China advances in the field of facial recognition privacy protection
legislation. At the same time, we should also note that laws and regulations on
personal data protection have lagged behind reality, and laws are separate from each
other and have lower legal effects. The bright side is that the Data Security Law,
the Personal Information Protection Law, the Telecommunications Law, and the
Cryptography Law have been incorporated into the legislative plan.

74See Zhou Kunlin, Li Yue, Research On The Legal Regulation Of Face Data Use Under Response-
Based Theory, Southwest Finance, No. 12, 2019.
164 Y. Cui

3.3.4 The Legislation of Privacy Protection in Facial Recognition

in Other Countries

Legislation on facial recognition varies from country to country. In Canada, facial

recognition technology has been used by police departments. However, there is a
lack of specific regulations on the technology in the Personal Information Protection
and Electronic Documents Act and other privacy laws.75 In Japan, the law stipu-
lates that companies that meet certain criteria could collect consumer information in
accordance with the law. In general, despite different legislative progress in different
countries, overall, there has been rapid development in the field.

3.4 The Privacy Protection Law in the Application of Facial

Recognition Technology in China Needs Urgent
Improvement

Facial recognition technology is widely used in China and will move toward deeper
application in the future. The further integration of facial recognition technology and
machine learning, big data mining and other artificial intelligence technologies has
promoted the development and application of the technology. However, regulations
lag behind constantly evolving technology and lack a specific adjustment mechanism.
Therefore, worries and concerns about privacy protection in the application of facial
recognition technology are on the rise, and an increasing number of voices call for
special legislation in the field.

3.4.1 Strengthening the Awareness of Personal Information Privacy

Protection in Facial Recognition Technology Application

As facial recognition technology advances and is more widely used, the tension
between technology and privacy rights increases. In the current technological envi-
ronment, personal privacy is prone to be infringed upon. Therefore, there is a high
need to establish the concept of personal information privacy protection and raise
awareness of personal information privacy protection. In addition, from the perspec-
tive of legal structure, regulations and guidance on technologies related to facial
recognition are also needed so that the technology can benefit society.
Individuals are the owners and generators of face information and have the closest
relationship with their face information. From the perspective of individual subjects,
face information has obvious private attributes, and face information leakage will
cause great harm to the person concerned. Therefore, efforts should be made to
improve individuals’ awareness of face information as private attributes and foster

75See Stop Facial Recognition in Canada. Retrieved from https://act.openmedia.org/StopFacialRe

cognition. Last visited 22 April, 2020.
5 Special Reports on Development of Artificial Intelligence … 165

their awareness of privacy protection. Meanwhile, awareness of individual privacy

protection should be developed for organizations that develop facial recognition
technology.

3.4.2 Improving the Legal System of Privacy Protection in Facial

Recognition Technology Application

With the rapid development of personal digital information such as facial recognition,
individuals have gradually realized the importance of privacy protection. At the same
time, laws and regulations on privacy protection in facial recognition need further
improvement.

Improving Laws and Regulations on Privacy Protection in Facial Recognition

First, laws and regulations on personal information protection should be improved,

for example, establishing the ownership, use and license rights of personal image
data and establishing a relief and litigation mechanism for individuals’ informa-
tion rights. Second, the legislation on big data should be improved, norms for safe
collection, transmission and use of data from enterprises and the Internet should
be established, and the legal responsibilities of data collectors, storers and utilizers
should be implemented. Third, the application and management norms of biometric
information should be improved, and facial recognition data should be used in a safe
and compliant way. Fourth, implementing the Cybersecurity Law of the People’s
Republic of China, accelerating the formulation of the Personal Information Protec-
tion Law and other relevant laws and regulations, establishing the mechanism that
covers both facial recognition technology development and privacy protection.

Clarifying the Legal Responsibilities of Face Information Collectors

and Storers in the Legislation

As a type of sensitive personal information, a strict norm shall be followed in the

collection and use of facial recognition data.76 In other words, the collection of
facial recognition information should be performed for proper reasons to reduce
the improper use of such information. For government and other agencies, informa-
tion that is collected and stored for public purposes shall be used for public interest
only and should establish supporting systems such as information registration, use
filing, and information confidentiality. For business entities, the information security

76 See Yang Jian: Legal regulation of Facial Recognition technology application. Retrieved from
http://www.cssn.cn/zx/bwyc/202003/t20200325_5105433.shtml?COLLCC=2158995962&.
166 Y. Cui

responsibility principle shall be established, and take the responsibility of data secu-
rity by informing the purpose of information collection, the authority of data transfer
and sharing, and the security and remedial measures of the data bank.

Establishing the Legal Remedy Mechanism for Improper Information Use

To reduce harm caused by information leakage, legal remedies for information misuse
should be established. The tort remedy of improper use of personal information
shall be applied according to the tort liability law and other legal provisions. Group
facial recognition data leakage, for example, the leakage of information transmission
channels or large-scale personal face information leakage caused by data vulnera-
bility, should be applied according to administrative regulations, and security checks
should be performed for the database involved. In the case of large-scale data leakage,
administrative penalties can be imposed.

3.4.3 Improving Relevant Standards of Privacy Protection

in the Application of Facial Recognition Technology

Since there is a lack of legal guidance from the top level and common standards,
for the more reliable and controllable application of facial recognition technology,
standards in subfields such as the fields of science and technology, industry and trade
should be improved.

Establishing Facial Recognition Technology Application Standards for State,

Local and Industrial Levels

At present, clear industry access norms are needed in face recognition in China to
guide the development of face recognition technology. Meanwhile, apart from formu-
lating relevant technical standards such as general standards, application interfaces
and technical specifications, legal issues such as privacy rights protection, personal
privacy and database security should also be taken into consideration.

Promoting the Upgrade of Privacy Protection Technology and Its

Industrialization

Due to the lack of legal guidance from the top level and the lack of corresponding
safety standards in the industry, enterprises pay little attention to the development and
application of privacy protection technologies. In addition to the fact that the develop-
ment of such technology may increase the operating costs of enterprises, the illegal
consequences of improper use of such technology are relatively small; therefore,
enterprises tend to ignore the risk of face information leakage. Relevant authorities
5 Special Reports on Development of Artificial Intelligence … 167

should encourage and support the development of privacy-protecting technologies

to further promote the development of the industry.

Strengthening the Security Responsibility in the Application of Facial

Recognition Technology

In facial recognition applications, a safe and reliable path for the collection, use,
transmission and secondary use of face information should be constructed; when
information is used for sensitive purposes, evaluation and record filling are needed.
Clear industry access standards should be established to avoid the risk of defective
products and privacy leakage in facial recognition technology to reduce the improper
use of face information and improve the self-discipline of the industry.77

4 Research on Legal Issues Related to the Application of AI

in the Medical Field78

4.1 Overview

Healthcare is one of the areas where AI is most deeply applied. The first integration
of AI and healthcare dates back to 1972, when AAPHelp (Leeds University, UK), a
clinical decision support system, was introduced to help diagnose abdominal pain and
support operations. In 1974, The University of Pittsburgh launched the INTERNIST-
I system, designed to assist the diagnosis of complex diseases in internal medicine.
In 1976, Edward Shortliffe released MYCIN, an AI medical consultation program
for testing blood infections at the Stanford lab. The first use of a robot-assisted
surgical procedure in recorded history occurred in 1985 when the PUMA 560 robotic
surgical arm was used in a delicate neurosurgical biopsy, a nonlaparoscopic surgery.
According to the 2018 World AI Industry Development Blue Book by Gartner and
CAICT, AI + healthcare ranks among the hottest areas of AI + vertical application
in China, accounting for the largest proportion of 22%. Marching into the twenty-
first century, AI has changed the mode of diagnosis, data processing methods, health
management and many other aspects, providing smart, targeted and efficient modern
healthcare services. AI has been used in patient triage platforms, imaging diagnosis
and payments, which not only improved hospital efficiency but also advanced the
healthcare system.
In 2007, the DeepQA research team from IBM launched Watson, a computer
system that gave diagnosis and treatment advice fast enough by simply asking

77 See Jiang Jie, Infringement and Risk Control in the Application of Facial Recognition Technology,
Books and Information, Issue 5, 2019.
78 Wu Tao, researcher, Shanghai Judicial Think Tank Association; Zhao Yucheng, deputy director,

Shanghai Education Newspaper.

168 Y. Cui

medical history and symptoms from patents and analyzing natural language ques-
tions. In 2017, the U.S. Food and Drug Administration (FDA) approved CardioDL,
an AI software for analyzing cardiac MRIs. In 2018, IDx-DR became the first ever
autonomous AI system cleared by the FDA to provide a diagnostic decision. The
AI “Smart Medical Assistant” robot, jointly developed by IFlytek Co., Ltd. and
Tsinghua University, passed the 2017 Examination for Clinical Practitioners, and
the examination result was intermediate to advanced.
Entering the twenty-first century, China has implemented several policies on the
AI medical sector. For example, in 2008, the Eleventh Five-Year Plan for Health
Informatization Construction in Zhejiang Province issued by the Health Commission
of Zhejiang Province mentioned that “information technology such as AI, signal
processing and data mining should be widely used in Chinese medicine diagnosis
and biomedical engineering field”, and that was the first time that application of AI
in medical field has been included in a provincial policy document. In 2018, Anhui
Province issued the Development Plan of the New Generation Artificial Intelligence
Industry in Anhui Province (2018–2030) and Implementation Opinions of the General
Office of the People’s Government of Anhui Province on Promoting the Development
of “Internet + Medical Health”, which actively promoted the application of AI in the
medical field, including surgically assisted robots, medical image-assisted diagnosis
systems and medical image recognition systems. In 2018, the Shanghai Municipal
Government released Healthy Shanghai 2030 and the Opinions of the Shanghai
Municipal People’s Government on Promoting the High-Quality Development of
the City’s Health Service Industry and Accelerating the Construction of the City of
First-class Medical Center, which significantly promoted the application of AI in
the medical field. In September 2018, Beijing released the Beijing Action Plan for
Accelerating Collaborative Innovation in Medicine and Health (2018–2020), which
focused on the integration of healthcare and AI as well as big data technology and
cultivating new business models.79
The development of China’s AI applications in the medical field has also been
at the forefront of the world. Since 2018, the Institute of Artificial Intelligence of
Shanghai Jiao Tong University, Shanghai Health Development Research Center, and
Shanghai Jiao Tong University School of Medicine have jointly published the annual
AI Healthcare White Paper.80 In 2019, CHINC released the Annual Report on the
Application of AI in Medical Fields in China. In the field of technology application,
several medical image big data and image recognition technology-assisted diag-
nosis and treatment systems developed by Shanghai Jiao Tong University School of
Medicine have been applied to clinical practice in Shanghai Jiao Tong University
School of Medicine and its affiliated hospitals and have achieved good results.81 Since

79 See Hu Kehui, Chen Xiaoyun, etc.: Analysis of Provincial Medical Artificial Intelligence Policies
in China, China Digital Medicine, Volume 14, Issue 3, 2019.
80 See Yi Rong: Artificial Intelligence Medical Imaging Thriving, Shanghai Jiao Tong University

Releases the First AI Medical “Guide”, Xinmin Evening News, January 10, 2019.
81 See Wang Zilong, Jiao Shengyin: AI Fundus Screening Promotes AI Healthcare Implementation,

Zhangjiang Science and Technology Review, Issue 2, 2020.

5 Special Reports on Development of Artificial Intelligence … 169

2016, Peking Union Medical College Hospital has applied AI on fundus examination
and has made progress in diabetic retinopathy treatment.82
Countless data have been generated in the medical field, from medical imaging,
patient consultation, and medical behavior to data collected in drug development,
such as omics data, drug target data, and literature data. Relying solely on physicians,
hospital administrators, and pharmaceutical R&D staff to handle these data can result
in wasted human resources and inefficient processing. AI technology can be a helping
hand in data processing.
AI technology has four components: cognition, prediction, decision making and
integrated solutions. In the medical sector, AI is mainly used in smart imaging, smart
pathology, and smart decision making.83 From the perspective of legal behavior, the
legal issues involved can be divided into five main areas: medical data utilization,
medical device use, drug development, medical diagnosis, and medical institution
operation. However, current policy lags far behind technology advances. Due to the
lack of safety assessment systems and quality assurance systems, the public has
doubts about the legitimacy of products. For example, while the wearable devices
market is thriving, as of 2020, the country has not yet issued technical specifica-
tions for the production of medical wearable device sensors, the key component of
wearable devices.84

4.2 Legal Issues Related to the Utilization of AI Medical Data

4.2.1 Issues of Medical Data Protection

China’s existing legal framework has relatively perfect provisions for protecting
patients’ right to health and privacy. The Artificial Intelligence Assisted Therapy
Technology Management Norms issued by the Ministry of Health released its 2017
edition.
Data-based medical technology visualizes diseases, and for healthcare workers,
each patient becomes a unique individual instead of homogeneous data. At the same
time, data security risks are on the rise. In the network society, internet users’ aware-
ness of protecting personal data lags behind the development of the network, and
enterprises’ protection of network data security is not enough.85
For example, in 2019, in the product liability case between Inner Mongolia Frui
Medical Technology Co., Ltd. and Wuxi Heskel Medical Technology Co., Ltd.,

82 See Chen Lisheng, Zhang Zhengling: A Discussion on the Opportunities and Challenges of
Medical Services in Artificial Intelligence age, China Medical Device Information, January 2020.
83 See Cao Hui, Gu Jiayi: Challenges, Opportunities, and Thinking of Surgeons in the Artificial

Intelligence Age, Chinese Journal of Practical Surgery, Issue 1, 2018.

84 See Tan Zaixiang, Han Xiaoping, Ding Tiantian: The Dilemma and Solutions of Medical Artificial

Intelligence Application in China, Health Economics Research, June 2020, Vol. 37, Issue 6.
85 See Tang Kuiyu, Zhang Xu: The Digital Basis of Network Society Quality-Network Society

Evaluation of Small Data to Big Data, Journal of Dialectics of Nature, Issue 8, 2018.
170 Y. Cui

Beijing China-Japan Friendship Hospital was accused of using products involved

in the case, with a 4G Internet card. During the trial, the court found that patient
information may have been leaked because those products were used.

4.2.2 Issues of Medical Data Security

Data in the medical field need to be organized and labeled by personnel, and the data
acquisition process often requires many human, material and financial resources and
consumes considerable time. In addition, in modern society, data have become a new
type of production material, which has the characteristic of easy leakage. Hackers
may obtain data directly through web crawlers and other techniques.
In China, there have been data breaches and human genome data leaks because
of the use of patient data for AI training. Most existing laws and regulations were
about data security and individual privacy protection.86 For example, regulations
on medical data can be found in Article 1226 of the Civil Code of the People’s
Republic of China, Article 49 of the Law of the People’s Republic of China on the
Promotion of Basic Medical and Health Care, and Article 23 of the Measures for
the Administration of Internet Hospitals (for Trial Implementation). However, most
of the security regulations for medical big data are for the regulation of medical
institutions, and there is a lack of basic institutional norms and standards for data
development and utilization by medical big data companies, which requires further
refinement of management rules.
Data security management involves data collection, data storage, data usage, data
transmission, data sharing, etc. With the advent of the big data era, the current intel-
lectual property law, anti-unfair competition law and similar protection systems, as
well as contractual arrangements and technical measures, can no longer fully meet
the data protection needs. Sectoral laws such as intellectual property can be combined
with practice to explore new models and methods of data protection.

4.3 Legal Issues Related to the Use of AI Medical Devices

Since 2014, under Regulations on Supervision and Administration of Medical

Devices, the China Food and Drug Administration (CFDA) has formulated and
revised 5 regulations, including the Administrative Measures for Registration of
Medical Devices, the Administrative Measures for Registration of In Vitro Diagnostic
Reagents, the Administrative Measures for Medical Device Manuals and Labels,
the Administrative Measures for Supervision and Administration of Medical Device
Production, and the Administrative Measures for Supervision and Administration of
Medical Device Operation, establishing a systematic regulatory system for medical

86 See Ji Ping, Guo Rui, etc.: Ethical Review and Legal Considerations of Medical Artificial
Intelligence Product Development, Medicine and Philosophy, Vol. 41, Issue 5, 2020.
5 Special Reports on Development of Artificial Intelligence … 171

devices. As of the end of 2019, there were 879 existing device regulations, including
715 official drafts and 164 requests for proposals. A total of 185 regulations were
issued in 2019, among which 103 official drafts and 82 requests for proposals.
AI technology has three main applications in the medical field. First, diagnostic
applications include AI-assisted film reading and AI-assisted diagnosis.87 Second,
therapeutic applications include implantable devices and medical robots, such as
da Vinci surgical systems in laparoscopic surgery and pancreatic necrosis debride-
ment surgery. Third, healthcare applications, including medical devices for recovery
and health management, such as wearable medical devices that monitor vital signs,
physical status, exercise duration, metabolic status, etc.

4.3.1 Issues of Medical Device Infringement

The integration of AI technology and medical devices brings a more precise diag-
nosis. However, defective medical devices can be very dangerous. In most cases, the
use of these devices relies on medical professionals. According to the Regulations on
the Supervision and Administration of Medical Devices, in China, medical devices
refer to AI medical devices, instruments, computer software, etc., such as the da
Vinci surgical system and ET Medical Brain. Diagnosis and treatment provided by
such AI medical devices are normally considered professional medical treatment,
which belongs to the category of medical technology.
Professor Fan Xianqun from the Ninth People’s Hospital of Shanghai Jiao tong
University School of Medicine believes that medical institutions and physicians who
use AI devices are the subject of medical conduct and legal liability in AI-assisted
medical treatment. In China, valid cases of medical disputes concerning the da Vinci
surgical system were considered medical malpractice disputes, and the Liability for
Medical Malpractice in the Tort Law of the People’s Republic of China was applied
to judge the cases. The status of the da Vinci Surgical Robot as a special AI medical
product is not recognized. The damage caused by the AI product itself cannot be
fully classified as medical malpractice liability88 since causes of harm include not
only medical negligence and defective medical devices but also factors such as the
patient’s idiosyncratic constitution and natural variations of diseases. The diversity
of causes of harm also leads to difficulties in proving product defects. Taking the
responsibility of medical institutions as the starting point and the cause of harm as the
criterion, the harm caused by AI medical devices can be classified into the following
categories.89

87 See Liu Jianli: Legal Challenges and Responses to the Clinical Application of Medical Artificial
Intelligence, Oriental Law, Issue 5, 2019.
88 See Ding Lu: Analysis of AI Medical malpractice Liability—Taking da Vinci Surgical System

as an Example, Journal of Beijing University of Chemical Technology (Social Science Edition),

Issue 1, 2020.
89 See Cui Zhijun: On the Infringement Liability of Implantable Medical Devices, Journal of Hubei

Radio and TV University, 2016, Issue 4.

172 Y. Cui

First, harm caused solely by medical device defects. In this situation, patient harm
was caused solely by the defective AI medical device, and the medical institution was
not negligent in the selection and use of that medical device. This type of liability
is not a real joint and several liability, and if the medical institution is liable for
damages, it has the right to recover from the manufacturer or seller of the defective
medical device if it proves that it is not at fault for the patient’s damage.
Second, harm was caused solely by inappropriate use of medical devices. In this
situation, patient harm is caused by medical personnel’s inability to master the use of
such medical devices or some inappropriate operations of AI medical devices. This
type of liability is general medical tort liability, under the category of expert tort, and
does not involve product liability.
Third, harm caused by multiple factors. In this situation, patient harm is caused
by both AI medical device defects and medical malpractice of medical institutions.
Such harm appears to be more complex and diverse in form. This type of liability is
joint and has several liabilities, with the medical institution and the manufacturer or
seller of the AI medical device being held jointly and severely liable.
Fourth, overtreatment and potential risks were caused by the abuse of AI in the
medical field. According to a public report on the website of a tertiary hospital in a
city, since the hospital introduced AI technology for lung nodule image recognition,
the number of patients receiving lung nodule screening has increased by 73% in
the past three years, with an 81% direct revenue increase, and the total revenue
of the hospital has increased by 100% combined with surgical revenue. However,
the distribution of the disease spectrum of NCDs in the region has not changed
significantly according to the health statistics yearbook of the region. In an NLST
(the National Lung Screening Trail) study, 96.4% of benign nodules were positive
in CT screening, and a high false positive rate was also seen in the “early diagnosis,
early treatment” program carried out in rural China. The high false positive rate may
lead to side effects such as overdiagnosis and overtreatment and result in increased
patient anxiety.90
Professor Fan believes that, according to Article 1223 of the Civil Code on Tort
Liability for Medical Damage, the judgment of damage caused by medical devices
follows the rule of no-fault liability. However, the detection and analysis of defects in
medical products is often insufficient. As a result, medical institutions are often liable
for compensation to the patient. Although tort liability also provides that the medical
institution is entitled to recover from the liable manufacturer after compensation, in
practice, it rarely happens.

4.3.2 Legal Issues Related to AI Medical Diagnosis

A computer-aided detection and diagnosis (CAD) system is a class of computer

systems that aims to assist in the detection and/or diagnosis of diseases through

90
See Li Ming, Li Yuxi, etc.: Discussion on Several Issues of Medical Artificial Intelligence Ethics,
Medical and Philosophy, Volume 40, Issue 21, 2019.
5 Special Reports on Development of Artificial Intelligence … 173

a “second opinion”. CAD systems are classified into two groups: computer-
aided detection (CADe) systems and computer-aided diagnosis (CADx) systems.
Currently, AI medical imaging robots are recognized as objects in civil law relation-
ships in the world’s major developed countries, and as medical devices with software
systems installed, they assist doctors in disease diagnosis and treatment, and the final
judgment must be made by doctors. In April 2018, the US Food and Drug Adminis-
tration approved the world’s first AI medical device IDx-DR for disease diagnosis. In
China, according to the New Version of The Medical Device Classification Catalog
released by China’s Food and Drug Administration (CFDA) on August 1, 2018, if
the diagnostic software provides diagnostic advice through its algorithm, instead of
directly giving diagnostic conclusions, it is a Class II medical device. If the software
automatically identifies the lesion and provides clear diagnostic tips, it is a Class III
medical device.
The general background of the commercial implementation of AI medical imaging
is the obvious shortage of imaging physicians in China. Each year, there is a 30%
increase in imaging examinations with only a 4% increase in imaging physicians. On
the one hand, such gaps may place great pressure on hospitals and physicians, and
physicians tend to experience fatigue in repetitive tasks such as reading CT scans.
On the other hand, some physicians are not qualified to generate the scan report. AI
medical imaging products have many benefits in the following medical practice: (1)
Diagnosis. To improve the detection rate of disease symptoms and reduce missed
diagnoses. Deadly diseases such as cancers can be detected in an early stage, which
will greatly improve patient survival rates and save money for the patient family.
(2) Treatment plans. The detailed analysis of lesion location and shape from AI is
helpful in preoperative and postoperative risk assessment. It is worth mentioning
that such technology is not yet mature enough. (3) Reading CT scans. AI medical
systems help to improve CT scan reading efficiency and save physicians’ time.91
The major causes of disputes about pathology are as follows: (1) Misdiagnosis.
False positive and false negative results; (2) missed diagnosis; (3) other notable
wrongdoings, such as specimens lost before the pathological examination or spec-
imen labeling errors. Currently, the final results of AI imaging still need to be manu-
ally verified and reviewed. Based on the preliminary diagnostic report of AI medical
images and his own expertise and experience, the imaging radiologist formed a
final diagnostic report for the clinician’s reference in confirming the diagnosis and
treatment. The clinician thus assumes responsibility for the patient’s diagnosis and
treatment results.
According to professor Fan, when determining whether there is a diagnosis fault,
the principle of determining the general diagnosis and treatment level should be
applied, that is, article 1224, paragraph 3, of the Civil Code of the People’s Republic of
China, which states that a medical institution shall not be liable to pay compensation
in the event that a patient has suffered damage in diagnosis and treatment if the
existing medical standards are not high enough to treat the patient.

91 See He Lianhong, Wang Zhixiong: Legal Issues of Compensation for Infringement in Artificial
Intelligence Medical Imaging, Politics and Law, Issue 3, 2020.
174 Y. Cui

4.3.3 Legal Issues of Consumer Protection

According to the definition of the right to consumer safety in Article 7 of Law of

the People’s Republic of China on Protection of Consumer Rights and Interests,
consumers shall, in their purchasing and using AI healthcare products, enjoy the
right of the inviolability of their personal and property safety. Consumers shall have
the right to require operators to provide AI healthcare products and services that meet
the requirements of personal and property safety. The subject of right refers to the
nature person who purchases, uses AI medical products or receives services from AI
medical devices; the content of right includes personal safety and property safety.92
Censorship and entry requirements are needed in traditional medical devices, and
censors must have the expertise qualified for the job. In regard to AI, the expertise
required of censors includes knowledge of neuroscience, mathematics, and computer
science, such as computational vision, natural language processing, and machine
learning.93
In the field of AI medical device consumption, due to the rapid iteration of tech-
nology and the lagging behind of relevant legal systems, there are still problems such
as imperfect access mechanisms of early products, imperfect risk control mecha-
nisms of mid-term products and imperfect relief mechanisms of defects in products.
Alongside the thriving market, censorship in AI medical products remains a weak
link.
According to the New Version of the Medical Device Classification Catalog
released by CFDA in China, which classifies medical assistance software based
on their potential risks, most AI medical products are Class III medical devices since
they have higher potential risks. As of 2020, no domestic company has obtained the
registration certificate of Class III medical devices from CFDA for its new generation
of AI products.
Unlike traditional medical devices, the function and algorithms of AI products
in the medical field were updated from time to time. For that reason, the traditional
approval process lags behind the rapidly developing industry. The Medical Device
Regulation (Decree 650) has classified healthcare software as medical devices. In
China, the main regulations on medical software are: the Instructions on the Basic
Requirements for Registration and Declaration of Medical Device Software, which
sets the requirements for software registration without touching on product secu-
rity and evaluation; the Supervision and Administration of Medical Device Soft-
ware Products in Beijing (Provisional), which makes clear provisions on issues such
as the production license of medical device software products in Beijing, without
providing detailed regulations on products classification, quality assessment and
product review, etc.

92 See Hu Yuancong, Lin Lixing: Discussion on the Consumer Safety Right of Artificial Intelligence
Medical Device Products, Journal of Nantong University, Issue 2, 2020.
93 See Li Yuanyuan: Artificial Intelligence Brings Innovative Development, Smart China, Issue 9,

2019.
5 Special Reports on Development of Artificial Intelligence … 175

In January 2012, the European Commission published document MEDDEV 2.1/6,

which provides guidance on the identification and classification of standalone soft-
ware in medical device regulatory systems. The U.S. made clear that traditional
medical device approval should not plague the application of AI in medical field
t. In July 2017, the U.S. The Food and Drug Administration released the Digital
Health Innovation Action Plan. The FDA’s Center for Devices and Radiological
Health (CDRH) aims to advance digital health by issuing guidance for medical soft-
ware products. The Pre-Cert for Software Pilot Program,94 as outlined in FDA’s
Digital Health Innovation Action Plan, aims to look first at the software developer or
digital health technology developer, rather than primarily at the product. The FDA
is currently basing the Pilot Program’s criteria on five excellence principles: patient
safety, product quality, clinical responsibility, cybersecurity responsibility, and proac-
tive culture. This new approach will greatly accelerate the approval process before
product release so that the latest technology can be applied as soon as possible.
Approval can also shed light on AI medical device approval in China.

4.4 Legal Issues Related to AI Medical Drug Discovery

and Development

AI technology helps new chemical drug development. Many cases can be found
in the overseas drug discovery market. For example, IBM and Pfizer reached an
agreement on developing immuno-oncology drugs in December 2016. IBM’s “Wat-
son” is conducting virtual drug screening, which is a computer simulation of the
drug screening process to predict and rapidly analyze the possible activities of drug
analysis, pharmacological toxicity, etc., and use a large amount of laboratory data,
clinical reports and scientific publications to test conjectures and effectively screen
the molecules of compounds that could be manufactured into drugs, to find potential
drugs and significantly reduce the cost of drug development.
In recent years, Chinese pharmaceutical companies such as Jiangsu Hengrui
Medicine Co. Ltd., Zhejiang Hisun Pharmaceutical Co. Ltd., and Shanghai Fosun
Pharmaceutical Co., Ltd., are increasing their investment in drug discovery and
development, with more attention given to complex diseases such as oncology. The
booming innovative drug market provides a sound environment for the application
of AI technology in the industry. In some cases, the use of AI has reduced 80% of
time in a single development session. China’s “AI+” pharmaceutical enterprises have
scattered business directions, which are vertically subdivided, and mainly enter the
supplier system of pharmaceutical companies through blind testing and academic
research cooperation. Currently, most of the services are provided to pharmaceu-
tical companies on a project basis. In the future, some AI enterprises will enter the

94U.S. Food and Drug Administration. Digital Health Software Precertification Program[EB/OL].
[2018–09-07]. https://www.fda.gov/medicaldevices/digitalhealth/digitalhealthprecertprogram/def
ault.htm.
176 Y. Cui

research and development of their own new drugs to obtain more lucrative profits,
which may lead to the emergence of new research and development entities.
With the help of big data and AI technology, clinical trials may no longer be
needed in drug discovery, which may lead to illegal activities.95
Unlike Industry-Sponsored Clinical Trials (ISTs), which were conducted by phar-
maceutical companies, investigator-initiated clinical trials (IITs) are clinical drug
studies initiated by researchers for the purpose of improving clinical treatment effi-
ciency. According to Article 2 of the Measures for the Management of Clinical
Research Projects Conducted by Medical and Health Institutions (Measures for the
Management of Clinical Research) issued by the National Health Commission of
China, IIT refers to all medical research involving human drugs conducted in health
care institutions. IIT, together with medical device research and the clinical applica-
tion of new technologies, was included in clinical studies conducted in health care
institutions. IST, as defined by Article 68, paragraph 2 of the Code of Practice for
Quality Management of Pharmaceutical Clinical Trials (2003), refers to any system-
atic study of a drug on the human body (patients or healthy volunteers) to confirm
or reveal the action, adverse effects and/or absorption, distribution, metabolism and
excretion of the test drug. IST aims to ensure that the testing drug is effective and
safe; it is also a legal procedure for the application registration of new drugs under
drug registration regulations.
The promulgation of Measures for the Management of Clinical Research was
based on the Law On Scientific and Technological Progress, the Law on Medical
Practitioners, the Law on Drug Administration, the Regulations on the Administra-
tion of Medical Institutions, and Nine Prohibited Acts in Good Medical Practice.
In content, apart from organizational management, project management, implemen-
tation management, and supervision and management, the Measures for Clinical
Research Management also includes a separate chapter on financial management,
which clearly requires medical and health institutions to take holistic management
of clinical research funds. In addition, clinics should set separate accounts for the
research projects mentioned above and implement separate accounting and dedicated
funds. Internal departments and individuals shall not receive clinical research project
funds, and recurring billing and illegal use of the funds are not allowed.

4.5 Legal Issues Related to the Operation of AI Medical

Institutions

Internet-based healthcare refers to online healthcare services such as online

consulting and diagnosis, teleconsulting, electronic prescribing, remote treatment
and rehabilitation. Broadly speaking, Internet medical services include core services
such as online diagnosis and treatment, electronic prescriptions, and supportive

95See Zhang Xudong: Report on China’s Medical Artificial Intelligence, Social Sciences Literature
Press, 2019 edition, p. 16.
5 Special Reports on Development of Artificial Intelligence … 177

services such as booking online appointments and online consultation. Those who
provide core services are known as Internet-based hospitals.

4.5.1 Market Access of Internet-Based Hospitals

First, medical entities must obtain a Medical Institution Practice License. According
to the Measures for the Administration of Internet-based Hospitals issued in July
2018, standalone Internet-based hospitals should be set up on the basis of medical
entities and shall not provide any medical services beyond the scope of those medical
entities.
Second, the doctors and nurses of Internet-based hospitals should be registered
medical practitioners.
Third, Internet-based hospitals must apply for and obtain an Internet content
provider (ICP) filing or business operation license. Only doctors registered in
the Internet-based hospital can carry out Internet medical treatment activities and
do not charge patients for the information they provide online, which is a self-
operated medical treatment activity. If the services provided by Internet-based hospi-
tals are nonoperating Internet information services, which offer public and shared
information to users through the Internet, then the hospital must keep an ICP record.
In medical partnerships, medical entities establish digital platforms by them-
selves or by cooperating with third parties. In this way, doctors originally registered
in other medical institutions can give online diagnoses through the platform. Prof-
itable Internet information service refers to the provision with charge of payment of
information through digital platforms to other medical institutions. According to the
Telecommunication Regulation of the People’s Republic of China and Administra-
tive Measures for Internet Information Services, medical entities shall apply for a
B-25 license, namely, a value-added telecommunication business license for Internet
information service, also known as an ICP license. Standalone Internet-based hospi-
tals, which not only charge medical institutions or doctors for information service but
also provide online digital processing service and online transactions between doctors
and patients, should also apply for a B-21 license, value-added telecommunications
business license for online digital processing service and online transactions.

4.5.2 Intellectual Property in Medical Institutions

AI medical care is a multidisciplinary, knowledge-intensive and capital-intensive

high-tech industry, and intellectual property protection is indispensable for both
product innovation and enterprise development. With the rapid development of the
medical device industry, there have been many cases of intellectual property disputes
in smart medical care in China in recent years.
Since AI technology is mainly based on algorithms and models to solve specific
problems, most of the innovative solutions of “AI + Medical” are combinations
of hardware and software or presented in the form of software only, and there are
178 Y. Cui

few innovative solutions involving hardware improvement only. Therefore, there are
still some uncertainties in the specific applicable standards of the relevant laws and
regulations of the Patent Law, the interpretation and determination of claims, and the
comparison between the features of the alleged infringing products and the patented
technical solutions.96
In legal practice, issues involving algorithms and models have always been a
difficult part of patent examination, and it is equally difficult to define whether
computer-assisted medical diagnosis and treatment solutions belong to disease treat-
ment methods. These difficulties are expected to be solved through more guiding
cases and are also expected to be standardized and unified through further revision
of the Patent Examination Guidelines. In addition, there are differences in the inter-
pretation and identification of computer software in patent examination, which has
been a long-standing controversy. There are no specific regulations and no practical
experience on how to interpret and identify the newly added storage media claims
in patent infringement litigation and whether they can play the expected role. It is
difficult to give a clear explanation and uniform regulations in a short time for these
disputes and confusions in the stage of confirming rights and safeguarding rights,
which still need to be studied and explored in the future.

4.6 Issues Related to AI and Healthcare Ethics

The ninth meeting of the Central Committee for Deepening Overall Reform in an
All-Around Way on July 24 approved the plan to establish a national committee
for ethics in science and technology. The National Medical Products Administra-
tion (NMPA) has issued guidance for the application of AI in the medical field,
providing more detailed regulations on the application of AI in medical field defi-
nitions, risks and clinical trials. In October 2019, NMPA issued an announcement
approving the establishment of a working group for the standardization of AI medical
devices. However, currently, there is no such provision for AI healthcare ethical
review. According to article 32 of the Law of the People’s Republic of China on
Basic Healthcare and Health Promotion, “When receiving medical services, citi-
zens shall have the right to informed consent for the illness, diagnosis and treatment
plans, medical risks, medical expenses and other matters according to the law”, and
“Clinical experiments and other medical research of drugs and medical devices shall
comply with the medical ethics and specifications, pass ethical review according
to the law, and obtain informed consent.” In medical practice, some hospitals have
adopted the ethical review system for clinical drug trails in AI medical ethical review,
and patients should sign an informed consent form before receiving AI healthcare
treatment to protect their right to informed consent and medical choice.

96
See Hong Yan: Analysis of Patent Protection of Artificial Intelligence Technology-Taking the
Medical Field as an Example, Intellectual Property, Issue 12, 2018.
5 Special Reports on Development of Artificial Intelligence … 179

Currently, giants in healthcare AI, such as IBM and Google, have issued product
principles and ethical regulations for AI products. AI Open-Scale from IBM, open-
source tools, and solutions from AI startups also helped greatly. The American
Medical Association and the Nuffield Council on Bioethics in the UK have proposed
ethical principles and considerations for the application and development of AI tech-
nology in healthcare. The guidelines from AMA particularly emphasized promoting
the development of thoughtfully designed, high-quality, clinically validated health
care AI that was (1) designed and evaluated in keeping with best practices in user-
centered design, particularly for physicians and other members of the health care
team; (2) transparent; (3) conforming to leading standards for reproducibility; (4)
identifying and taking steps to address bias and avoiding introducing or exacer-
bating health care disparities, including when testing or deploying new AI tools on
vulnerable populations; and (5) safeguarding patients’ and other individuals’ privacy
interests and preserving the security and integrity of personal information.
In China, some medical industry associations have established AI subcommit-
tees, advancing the application of AI in medical field ethical research. The Ultra-
sound Physicians Branch of the Chinese Medical Doctor Association has released
the Beijing Declaration on the Code of Conduct for Artificial Intelligence in Ultra-
sound Medicine in China, containing 13 rules and regulations in the following areas:
establishing norms and regulations for management; realizing medical-industrial
integration and promoting transformation; and maximizing patient benefits.97

4.7 Conclusion

China has long been plagued by the geographic imbalance in medicine resources, and
the application of AI technology in the healthcare and medical sectors offers better
solutions. The successful experience and achievements of China’s smart health care
can provide a series of important wisdom for the smart health care model. However,
while presenting opportunities for development, AI in the medical field also brings a
number of legal and ethical challenges. Therefore, to improve the medical level and
promote human health, we should make legal regulations and ethical guidance on the
application of AI in the medical field from the aspects of clarifying legal subjects,
dividing legal responsibilities, protecting data and privacy rights, and establishing
social norms and ethical guidelines.

97
Li Ming, Li Yuxi, etc.: Discussion on Several Issues of Medical Artificial Intelligence Ethics,
Medical and Philosophy, Volume 40, Issue 21, 2019.
180 Y. Cui

5 Research on Legal Issues Related to Blockchain

Technology98

In 2019, “blockchain technology+” started to enter the economic and social life of
Chinese people, and in 2020, we have seen its even rapid development. On October
24, 2019, the Political Bureau of the Communist Party of China (CPC) Central
Committee held its 18th group study on the development of blockchain technology.
During the meeting, President Xi Jinping, the general secretary of the CPC Central
Committee, stressed that we should take blockchain as an important breakthrough
in independent innovation of core technology, define the main direction, increase
investment, focus on conquering a number of key core technologies, and accelerate
the development of blockchain technology and industrial innovation. He also noted
that we should make the best of blockchain technology to promote data sharing,
optimize business processes, cut operating costs, improve collaboration efficiency,
and build a trustworthy network. Alongside the remarkable triumph, the technology
also brings about related legal issues.
Essentially, a blockchain is a set of new governing structures. It constructs a
“value network” and “trust network” through distributed data warehouses, peer-to-
peer transmission, consensus mechanisms and cryptographic algorithms. In this way,
network participants mount a collective effort to safeguard value and trust and are
devoted to forming a better governance framework, which empowers the modern-
ization of a country’s governance system and capacity. Blockchain technology, as
an important force in current productivity technological innovation, will change the
method of information production, flow and storage because of its characteristics
of being an immutable ledger. It will also shape digital assets, product traceability,
copyright protection, data sharing and many other areas. Currently, blockchain tech-
nology has been applied in digital finance, the Internet of Things, smart manu-
facturing, supply chain management, digital asset trading and many other fields.
Blockchain technology will have an important impact on both production relations
and superstructure today, especially on legal relations.
Blockchain technology and rule of law are new, and the development of healthy
blockchain technology and the rule of law complement each other. Only blockchain
technology based on the rule of law can help to improve the national governance
system and the modernization of governance capacity. Especially in today’s compre-
hensive deepening reform and comprehensive economic and social transformation,
it is of great importance to promote the healthy development of blockchain tech-
nology based on the rule of law. On the one hand, under the idea of “rule of law+”
blockchain technology, the formulation and implementation of relevant policies and

98Authors: Cao Hongxing, Director of Information Management Division, Shanghai Higher

People’s Court; Wu Haiyin, Deputy Director of Information Management Division, Shanghai Higher
People’s Court; Chen Jidong, Lecturer, Shanghai University Law School; Gong Sihan, Assistant
Researcher, Big Data and Artificial Intelligence Law Research Center, Shanghai University; Wang
Huachun, Doctoral Supervisor, School of Government Management, Beijing Normal University;
Zhao Chunxue, CEO of Suzhou Youxuan Intelligent Technology Co., Ltd.
5 Special Reports on Development of Artificial Intelligence … 181

laws will enable and promote the development of blockchain technology and indus-
trial innovation; on the other hand, the “blockchain technology+” rule of law means
that blockchain technology development will enable and promote the formulation
and implementation of relevant policies, plans and laws. This report discusses major
issues in blockchain technology and rule of law, including policies and regulations
on blockchain technology, jurisprudence of blockchain technology, analysis of the
application of blockchain technology in rule of law and typical cases of blockchain
technology. It is hoped that this report will promote the integration of blockchain
technology and rule of law and facilitate blockchain technology development as well
as industrial innovation in China.

5.1 Policies and Regulations on Blockchain Technology

As one of the most profitable new technologies, its potential has long been seen
and highly valued by some countries and will likely usher in huge dividends in
the future. Blockchain technology policies and regulations are closely related to
the technological development stage. A brief review of the development history of
blockchain shows that the development of blockchain technology has gone through
three stages. The first stage is “blockchain technology+” bitcoin, that is, blockchain
technology 1.0, and its corresponding policy planning and laws and regulations
are characterized by “strengthening supervision”. The second stage is “blockchain
technology+” finance, that is, blockchain technology 2.0. Its corresponding policy
planning and laws and regulations are characterized by “strengthening supervi-
sion+promoting standardization”, and its focus is on promoting standardization. The
third stage is “blockchain technology+” economic society, that is, blockchain tech-
nology 3.0, and its corresponding policy planning and laws and regulations are char-
acterized by “strengthening supervision management + promotion of standardization
+ promotion of development”, with a major focus on promoting development.

5.1.1 Policies and Regulations on Blockchain Technology in China

Overview

(1) Types. Analyzing at the national and local levels, there are three types of policies
and regulations on blockchain technology in China: policies and regulations
aimed at strengthening regulation, policies and regulations aimed at promoting
standardization and policies and regulations aimed at facilitating development.
(2) Time. At the national level, from 2013 to the end of April 2020, at the national
level, there were 51 policies and regulations related to blockchain technology,
including 1 in 2013, 2 in 2016, 7 in 2017, 9 in 2018, 19 in 2019 and 13 as of
the end of April 2020. From a macro perspective, the regulations fall into the
following three main categories.
182 Y. Cui

First, the introduction time of policies and regulations aimed at strengthening

regulation. Blockchain technology originated from bitcoin, a kind of cryptocurrency.
that circumvents the normal regulation of central banks and is used in large numbers in
gray and even criminal areas. Therefore, the state must strengthen the supervision of
digital currencies based on blockchain technology and relevant transactions. The laws
and regulations of China’s blockchain technology also start from supervision. With
the further expansion of the application of blockchain technology, the corresponding
policies and regulations are further enriched and improved. In December 2013, the
People’s Bank of China and five other ministries and commissions issued the Notice
on Preventing Bitcoin Risks, which was the first regulatory document in the field.
The number of policies and regulations introduced in 2017 was one; The number
introduced in 2018 was 4; In 2019, the number was 6, among which the Regulations
on the Management of Blockchain Information Services issued by the State Internet
Information Office in January was of great significance. By the end of April 2020,
the number of policies and regulations issued was one.
Second, the introduction time of policies and regulations aimed at promoting stan-
dardization. The number of policies and regulations issued in 2016 was 2, started by
the first White Paper on the Development of Blockchain Technology and Applications
in China issued by the Ministry of Industry and Information Technology in October;
the number of policies and regulations issued in 2017, 2018 and 2019 was 6 and 5,
5, respectively; In 2020, as of the end of April, the number was 1.
Third, the introduction time of policies and regulations aimed at facilitating devel-
opment. The number of policies and regulations issued in 2019 was 8, in October, the
State Council issued Guidelines on Deepening Reform and Ensuring Food, which
clearly promoted the application of blockchain technology in food security for the
first time and “blockchain+” technology has entered the economic and social fields
in an all-round way; In 2020, as of the end of April, the number was 11.
At the local level, policies and regulations were introduced relatively late, mainly
aimed at facilitating development. From 2016 to the end of April 2020, the total
number of policies and regulations was 163. The number of policies and regulations
issued in 2016 was 4. In August, Beijing Municipality introduced the Beijing Munic-
ipal Bureau of Financial Work 2016 Annual Performance Tasks, marking the start of
issuing policies and regulations on blockchain at the local level. In 2017, 2018 and
2019, the numbers were 31 and 43 and 56, respectively.
By the end of April 2020, 31 provinces, autonomous regions and municipalities
directly under the central government of China had introduced policies and regu-
lations on blockchain technology, among which Beijing, Guangdong, Shanghai,
Zhejiang and Jiangsu took the lead, followed by Hunan, Chongqing, Sichuan,
Fujian, Guizhou, Hainan, Yunnan, Anhui, Tianjin, Hebei, Guangxi, Gansu. Regions
with relatively backward economies, such as Qinghai and Tibet, also embarked on
introducing policies to support blockchain technology development.
5 Special Reports on Development of Artificial Intelligence … 183

Features

(1) National-level policies provide guidelines, and local-level policies provide

supplements. Policies at the national level have established the framework of
“strengthening regulation + promoting standardization + facilitating devel-
opment”, and policies at the local level are mainly aimed at facilitating the
implementation of national policies in the region.
(2) Current policies are mainly focused on promoting industrial development.
China’s blockchain technology has entered the 3.0 era, and promoting the
development of the blockchain technology industry has become the main
task at present. In line with the requirements of scientific and technolog-
ical development, the current blockchain policy mainly focuses on promoting
development.
(3) Local-level policies are positively correlated with the development level of
the local blockchain industry. The development of blockchain technology and
industry is closely related to policy support. Regions with a relatively advanced
development of blockchain, such as Beijing, Shanghai, Guangdong, Jiangsu,
Zhejiang and Shandong, issued more policies, while the number of policies in
Fujian, Hunan, Guizhou, Chongqing, Sichuan, Anhui, Hubei, Hainan, Yunnan
and Guangxi was also on the rise, reflecting the high importance that local
governments attached to the blockchain industry.

5.1.2 Policies and Regulations on Blockchain Technology in Other

Countries

Overview

The development of blockchain technology policies and regulations in other countries

is basically synchronized with the development of blockchain technology and has also
gone through the early stage of “strengthening regulation”, to the stage of “strength-
ening regulation + promoting standardization”, and then to the stage of “strength-
ening regulation + promoting standardization + promoting development”. Bitcoin’s
early payment methods were mainly online payments, and because of its anonymity
advantage, it was early adopted by illegal online traders, and without stronger regu-
lation, it could become an “online model of money laundering machine”. In March
2013, the U.S. The Financial Crimes Enforcement Network (FinCEN) issued the
Application of FinCEN’s Regulations to Persons Administering, Exchanging, or
Using Virtual Currencies, stipulating that Bitcoin transactions should be regulated
and taxed, marking a start of regulation of blockchain technology in the United States.
Bitcoin first gained its legal classification in Germany. In August 2013, the Federal
Ministry of Finance of Germany officially recognized bitcoin as a “monetary unit”
and a “private asset”. Germany is the first country in the world to recognize the legal
currency status of Bitcoin.
184 Y. Cui

(1) Time. Globally, policies and regulations were first launched in 2013. Devel-
oped countries such as the United States and Germany have successively issued
“blockchain + “bitcoin regulatory policies and started a new era of blockchain
technology regulatory policies and regulations. Bitcoin trading in the exchange
rate market should be regulated and pay transaction taxes.
From 2013 to 2015, since the issuance of regulations by FinCEN in March 2013,
countries around the world have successively issued a large number of blockchain
policies and regulations, which mainly focus on the “blockchain+” digital virtual
currency supervision field. FinCEN is a bureau of the United States Department of
the Treasury, and this is the first time that the US government launched policies on
distributed virtual currency.
Since the UK published distributed ledger technology, beyond blockchain in
January 2016, many countries have recognized the value of “blockchain+” finance
and introduced many policies and regulations aimed at promoting standardization
while regulating “blockchain+” cryptocurrency from 2016 to 2018.
Algorithms that enable the creation of distributed ledgers are powerful, disruptive
innovations that could transform the delivery of public and private services and
enhance productivity through a wide range of applications. A distributed ledger is
essentially an asset database that can be shared across a network of multiple sites,
geographies or institutions. All participants within a network can have their own
identical copy of the ledger. Any changes to the ledger are reflected in all copies in
minutes, or in some cases, seconds. The assets can be financial, legal, physical or
electronic. The security and accuracy of the assets stored in the ledger are maintained
cryptographically through the use of ‘keys’ and signatures to control who can do
what within the shared ledger. Entries can also be updated by one, some or all of the
participants, according to rules agreed upon by the network.
In July 2019, the US Congress introduced the Blockchain Promotion Act of 2019;
later, in September, Germany published a national blockchain strategy (Blockchain-
Strategie der Bundesregierung). Since then, an increasing number of countries have
noted the value of “blockchain+” in empowering socioeconomic development and
have issued a large number of policies and regulations to promote the develop-
ment of blockchain. The national blockchain strategy of Germany proposed specific
measures in five areas for action, including securing stability and stimulating innova-
tions in the financial sector; advancing innovation by providing funding for projects
and living laboratories; facilitating investment by putting reliable investment condi-
tions in place; and harnessing digital technology to improve administrative services
and distributing information and knowledge. According to Peter Altmaier, Federal
Minister for Economic Affairs and Energy, the national blockchain strategy aimed to
help maintain and expand German’s leading position in blockchain technology. One
focus of German blockchain technology is on the energy sector, and the national
plan will enhance the role that digital technology plays in the transition to clean
energy.” German Finance Minister Olaf Scholz stated that blockchain technology is
the cornerstone of the future Internet, and it is essential to protect consumers and
state sovereignty in the application of blockchain technology.
5 Special Reports on Development of Artificial Intelligence … 185

(2) Policy types. There are three types of policies on blockchain technology in
other countries: policies aimed at strengthening regulation, policies aimed at
promoting standardization and policies aimed at facilitating development. In
the past, policies were mainly aimed at strengthening regulation and promoting
standardization, and in recent years, policies aimed at facilitating development
were on the rise.
Policies aimed at strengthening regulation. Bitcoin ushered in Blockchain Tech-
nology 1.0, in which governments or central banks around the world have introduced
relevant policies and regulations on “blockchain technology+” virtual currency to
meet the requirements for developing virtual currency. The main object of those
policies was to include “blockchain technology+” virtual currency in a regulatory
framework.
Policies aimed at promoting standardization. With the advent of the blockchain
technology 2.0 era, countries all over the world have gradually realized the strategic
value of blockchain technology itself and its data assets and have mainly issued
relevant policies and regulations for “blockchain technology+” finance, the main
content of which is to standardize blockchain technology itself and its data assets.
With the advent of the blockchain technology 3.0 era, countries all over the world
have realized the strategic significance of blockchain technology in reconstructing
economic and social development and issued relevant policies and regulations mainly
for the “blockchain technology+” economy and society, the main content of which
is to promote the development of blockchain technology and industry.

Features

The general features of blockchain policies in countries and regions around the world
are described as follows:
(1) Emphasizing cryptocurrency regulation. In countries that recognize cryptocur-
rency as a legal tender, such as the United States, Germany and Japan, system-
atic laws and regulations that specifically deal with cryptocurrencies have been
issued. In countries that do not recognize cryptocurrency as a legal tender, such
as Russia, laws and regulations were issued to ban or restrict the use and trade
of cryptocurrency.
(2) Supporting the innovation of blockchain technology. The innovation and devel-
opment of blockchain technology calls for increasing resources and funds.
Attracted by its significant application value, many countries have acceler-
ated the layout of the blockchain industry. In the United States, many laws
and regulations of blockchain technology were passed in 2019, which empha-
size the application of the technology, such as allowing the government to use
blockchain technology.
(3) Improving policies and regulation. While blockchain technology is constantly
“empowering” the economy and society, it will certainly bring various risks,
such as legal issues. Hence, countries around the world have stepped up the
186 Y. Cui

formulation of relevant policies and regulations to build a blockchain rule of

law mechanism to make the most of blockchain technology.
(4) Establishing technical standards. With the rapid development of blockchain
technology, formulating unified technical standards has become the key to the
development of the industry. Therefore, an increasing number of countries have
accelerated blockchain technical standards and test benchmarks.
(5) Cultivating talents. The massive talent shortage has been plaguing the
blockchain community. More countries have increased investment in education
in this field to bridge the talent gap in the blockchain industry.

5.2 Legal Theoretical Research of Blockchain Technology

5.2.1 A Review of Key Areas of Rule of Law of Blockchain Technology

Research

With Blockchain Technology 3.0 era approaching, China chooses blockchain tech-
nology as an important breakthrough for independent innovation of core technolo-
gies.99 The legal profession has gradually deepened its understanding of the charac-
teristics and application scenarios of blockchain technology, yet core concepts and
theoretical framework of the legal research of blockchain technology have not been
formed. Therefore, sorting out problems and research progress in blockchain tech-
nology jurisprudence research lays the groundwork for understanding the technology
and making relevant regulations.
Research on the rule of law of blockchain technology focuses on three important
issues. First, the subject supervision of the blockchain network is mainly reflected
in the difficulty in defining the internal relationship of the blockchain caused by the
unique structure of blockchain technology. The second is the legal definition of smart
contract, the core of which lies in whether smart contract is a legal contract. Third,
the definition of cryptocurrency, that is, how to legally define encryption digital cash.
To address the above issues, scholars have shared their insights from the perspective
of legal theory and blockchain applications.

The Subject and Supervision of Blockchain Network

Consensus algorithms and automated execution of policies and rules create difficul-
ties in regulating blockchain technology. The blockchain system is invulnerable to
single point failure due to a trusted network, commission consensus and transparent
trading. Therefore, they are less insulated from the law’s reach than it seems. For
example, the Securities and Exchange Commission (SEC) of the United States issued

99 See Xi Jinping, “Taking Blockchain as an Important Breakthrough for Independent Innovation

in Core Technology”, People’s Daily (Overseas Edition) (October 26th, 2019).
5 Special Reports on Development of Artificial Intelligence … 187

an investigative report that found that tokens offered and sold by a “virtual” organi-
zation known as the “DAO” were securities and therefore subject to federal securities
laws.100 However, it takes considerable trouble to choose whether investors, sponsors
or third parties should bear illegal liability. Although the SEC ultimately chose issuers
or a third party as the subject of liability, this determination has caused considerable
controversy.101
DAO posted the question of how law should respond to the technology. Some
scholars gave their answers. Kevin Werbach viewed a blockchain network as a series
of concentric circles. In the middle is the ledger, secured through robust decentralized
consensus. In the next layer are the smart contracts, the software code that direct
transactions on the network. Outside are edge service providers such as exchanges
and wallet services, which interface between cryptocurrencies and the traditional
world. Finally, there are coins that DApps and others sell directly to users.102 Kevin’s
model explained the trust architecture of blockchain.
Professor Usha R. Rodrigues noted that DAO tokenholders, which were in the
outermost layer of the concentric circle, should be regulated. In this way, tokenholders
form partnerships through smart contracts. In this solution, the ownership interest
would not generate profits “solely through efforts of others.” Instead, it would look
more like a true partnership interest. The result would be that the DAO will function
more like the partnership under the law.103
However, Dr. Wang Qingsong believes that since blockchain has no obvious
internal layer architecture and the system automatically grants and disqualifies
members in a way different from traditional partnerships or companies, it is difficult to
regulate the subject of blockchain by the subject form of organization in the current
law. However, because the internal relationship of the blockchain system has the
characteristics of “team production” and “contract connection” and has a “common
purpose” due to the joint implementation of system services and “joint control”
between nodes, it can be proven that blockchain is a kind of “quasi-organization”,
that is, as a limited partnership; hence, corresponding laws and rules can be applied.104
In this regard, Kevin Werbach thinks that it is not so much how to regulate blockchain
as how to use blockchain technology to regulate it.105 The foundation of “managing

100 The DAO, U.S. Securities & Exchange Commission, (July 2017), https://www.sec.gov/litiga
tion/investreport/34-81207.pdf.
101 Jonathan Rohr and Aaron Wright, Blockchain-Based Token Sales, Initial Coin Offerings, and

the Democratization of Public Capital Markets, 37 (October 2017), https://papers.ssrn.com/sol3/

papers.cfm?abstract_id=3048104. Robinson Randolph, The New Digital Wild West: Regulating
the Explosion of Initial Coin Offerings (Sept. 1, 2017), https://papers.ssrn.com/sol3/papers.cfm?
abstract_id=3087541.
102 See Kevin Werbach, “Trust, However, Verify: Why the Blockchain Needs the Law”, translated

by Lin Shaohua, Shanghai People’S Publishing House, p. 44.

103 See Usha R. Rodrigues, “Law and the Blockchain”, Iowa Law Review, vol. 104, no. 2 (2019),

pp. 699—701.
104 See Wang Qingsong: Legal Science, Vol. 5, 2019.
105 See Kevin Werbach, “Trust, However, Verify: Why the Blockchain Needs the Law”, translated

by Lin Shaohua, Shanghai People’s Publishing House, p. 44.

188 Y. Cui

blockchain with blockchain” is Lesger’s viewpoint of “code is law”, according to

which blockchain technology itself is a governance tool. The integration regula-
tion mode of blockchain technology combined with law needs legal coding or code
legalization.

The Legal Characterization of Smart Contracts

Whether a smart contract is a legal contract is the primary topic of legal discussion,
and the issue remains highly debated among scholars. The negation holds that a
smart contract is just a specific name for the code that runs on and interacts with the
blockchain. It is triggered by a blockchain transaction and reads and writes data in
the blockchain’s database.106 Some scholars believe that smart contracts are neither
“intelligent” nor legal contracts and should not be included in the scope of contract
law.107 Dr. Jin Jing proposed earlier that smart contracts should be understood as a set
of performance mechanisms or execution procedures of an agreement because smart
contracts completely abandon classical theories or rules such as contract relief, legal
enforcement of contract disputes, interpretation of contract language, and contract
imperfection.108 Based on the self-executing property of computer automation, some
scholars have also defined smart contracts as a new form of “self-help behavior”.109
As to why smart contracts are not legal contracts, Dr. Wang Yanchuan further
explained that the core of traditional contracts is the will to pursue legal effects,
while smart contracts focus on the automatic execution of the will of the counter-
party. Smart contracts are a kind of “will-free” operation that eliminates people’s
will. At the same time, the will is not the execution of law but the execution of
machine software, so smart contracts are not legal contracts.110
Affirmation holds that although blockchain technology has built a brand-new
decentralized trust mechanism, it has not changed the contract essence of smart
contracts, and its structure still conforms to the “offer-promise” structure in tradi-
tional contracts and still belongs to legal contracts,111 but the expression of intent by
parties to a blockchain smart contract often has to be discovered by interpretation.112

106 See Adam J. Kolber, “Not-So-Smart Blockchain Contracts And Artificial Responsibility”,
Stanford Technology Law Review, vol. 21, no. 2 (Spring 2018), pp. 210—211.
107 See Daniel Cawrey, “Why Ethereum Needs ‘Dumb’ Contracts”, http://www.coindesk.com/eth

ereum-dao-dumb-smart-contracts/.
108 See Jin Jing, “The Power of Classical Contract Law in the Digital Age—In the Context of the

EU Digital Single Market Policy,” European Studies, Vol. 6, 2017.

109 See Max Raskin, “The Law and Legality of Smart Contracts”, Georgetown Law Technology

Review, no. 1(2017), p. 304.

110 See Wang Yanchuan, “The Construction of Smart Contracts and Risk Prevention,” Journal of

Law, Vol. 2, 2019.

111 See Zhao Lei and Sun Qi, “Smart Contracts in the Perspective of Private Law System,” Economic

and Trade Law Review, Vol. 3, 2019.

112 See Guo Shaofei, “Contract Law Analysis of Blockchain Smart Contracts,” Eastern Jurispru-

dence, Vol. 3, 2019.

5 Special Reports on Development of Artificial Intelligence … 189

Many scholars agree that the smart contract is a contract. Dr. Guo Shaofei thinks that
the parties to the smart contract agree to transfer their legal rights and obligations
through the smart contract, which is no less than the actual exchange of goods and
payment between the two parties through the counter.113 Dr. Cai Yibo further pointed
out that smart contracts are homogeneous with traditional paper contracts and elec-
tronic contracts, and the difference between them is mainly carrier problems, that
is, the carrier of smart contracts is code. However, smart contract technology deter-
mines the particularity of the carrier, the formation mechanism and the specific appli-
cable rules of intention expression.114 Dr. Jin Jing pointed out that smart contracts
are incomplete contracts; that is, self-executing systems and anonymity constitute
typical features of smart contracts, and the execution of smart contracts is closed,
which makes it possible for smart contracts to escape incomplete contract theory.115
Other scholars believe that a smart contract is nothing more than the encoding or
digital memory of a contract or part of it, and its legal evaluation depends on the laws
applicable to the underlying contract.116

Defining Cryptocurrencies

Blockchain technology will remain the underlying technology of cryptocurrency for

a long time. While cryptocurrency has triggered a series of changes in the financial
field, it has also brought a series of problems, such as unclear legal definitions,
financial crimes and regulatory difficulties.
For the legal definition of cryptocurrency, there are three main viewpoints: affir-
mative, negative and compromise. The core debate on whether cryptocurrencies can
be currencies is about “legal compensation”, that is, whether national credit is an
inherent attribute of currencies. Affirmative scholars believe that “legal compensa-
tion” is not the essential feature of currency, and the government should not be the
only issuer of currency. Cryptocurrency has the general characteristics that it should
have as currency, and its issuing and circulating mechanism is the same as that of
legal tender. Therefore, cryptocurrency should be recognized as parallel payment
currency in the current monetary system as a supplement to the existing monetary
payment system.117 Scholars who deny the property of cryptocurrency mainly have
two reasons. First, currency is essentially a circulation means supported by national

113 See Kevin Werbach and Nicolas Cornell, “Contracts Ex Machina”, Duke Law Journal, vol. 67,
no. 2 (2017), p. 313.
114 See Cai Yibo, “Study on the Fitting of Smart Contracts and Private Law Systems,” Eastern

Jurisprudence, Vol. 2, 2019.

115 See Jin Jing, “The Power of Classical Contract Law in the Digital Age—In the Context of the

EU Digital Single Market Policy,” European Studies, Vol. 6, 2017.

116 See Markus Kaulartz and Jorn Heckmann, “Smart Contracts-Anwendungen der Blockchain-

Technologie” (2016) CR 618, 622.

117 See Lou Yaoxiong and Wu Jun, “Analysis of Bitcoin Legal Issues,” Journal of Beijing University

of Posts and Telecommunications (Social Science Edition), Vol. 4, 2013.

190 Y. Cui

credit. Due to the lack of national credit support, cryptocurrency has no legal compen-
sation, and it is difficult to perform the function of commodity exchange as a stan-
dard currency.118 Second, due to the characteristics of decentralization, there is no
national credit endorsement, and due to the characteristics of limited total amount
and anonymity of transaction, the value of cryptocurrencies such as Bitcoin fluctu-
ates greatly, and they only play the role of circulation means in a limited range, so
they are not recognized as currencies.119 Scholars who take the “compromise view”
believe that cryptocurrencies have both the properties of “currency” and “financial
instrument”. When cryptocurrency serves as a consideration for the exchange of
goods and services, it acts as a transaction intermediary and a store of value, which
constitutes currency in a substantial sense. When cryptocurrency and fiat currency are
traded on a specific platform, it bears the functions of financial instruments for cash,
investment and exchange of fiat currency and constitutes financial commodities.120
Since 2014, the People’s Bank of China has started its study on legal digital
currency. The research and development of digital RMB has been advancing steadily.
Under the premise of two-tier operation, M0 substitution and controllable anonymity,
the digital RMB system has basically completed top-level design, standard formula-
tion, functional research and development, joint research and testing, etc. Following
the principle of being steady, safe, controllable, innovative and practical, the digital
RMB will be tested internally in Shenzhen, Suzhou, Xiongan New Area, Chengdu
and the future Winter Olympics to continuously optimize and improve its function.
The digital currency to be launched by the People’s Bank of China is different from
Bitcoin or Libra. Digital RMB is endorsed by national credit, which can be said to be
an electronic version of RMB. This move also shows China’s attitude toward digital
currencies.

5.2.2 Possible Directions for Future Legal Research on Blockchain

Technology

Technology Logic Must Be Emphasized in Legal Hermeneutics

on Blockchain Technology

In the future, the main focus of legal hermeneutics on blockchain technology is

to gain a deeper and more comprehensive understanding of blockchain technical
architecture and accommodate the technology into the current legal hermeneutics
system. Smart contracts and data rights may draw heated discussion in the field.
Smart contracts and data rights are likely to be the focus of future discussions.

118 See Sheng Songcheng and Zhang Xuan, “Virtual currencies are not currencies by nature—Take
Bitcoin as an example”, China Finance, Vol. 1, 2014.
119 See Wang, Su-Zhen, “Bitcoin from the Essence of Money,” China Finance, Vol. 9, 2014.
120 See Xie Jie and Zhang Jian, “The Choice of Economic Criminal Law in the Era of “Decentral-

ized” Digital Payment: A Legal and Economic Analysis Based on Bitcoin”, Legal Science, Vol. 8,
2014.
5 Special Reports on Development of Artificial Intelligence … 191

Taking the smart contract as an example, it has explicit limitations compared with
traditional contracts: weak flexibility, formalized obligations and issues related to the
standards of the contracts.121 We must find a way to properly identify the intention
of both parties and better manifest their willingness.

The Integration of Autonomous Regulating and Legal Regulating

Although autonomous regulation remains a dominant power in regulating blockchain

technology, legal regulation should not be omitted, especially in increasing multi-
stakeholder participation and constructing a global mode based on the consensus
about the application layer in the Internet protocol suite, deeper transport, network
architecture and blockchain technology architecture. More specifically, legal regu-
lation must strike a balance among blockchain technology, capacity and neolib-
eral ideology to potential technical bias and data hegemony of code and capital.
However, how to realize the codification of law, especially in a more scientific form,
requires a deeper understanding of the development of science and technology by
legal professionals and experts.122

Neoliberal ideology—The Foundation of Legal Research

Some scholars have proposed that both the blockchain technology ecosystem and
neoliberalism represent signs of “Tech for Social Good”. However, could blockchain
technology actually be used “for social good”? We could escape from central-
ized institutions and government authorities at the cost of being trapped in “code
sovereignty”. As Dr. Zheng Ge said, while embracing the new technology, we should
avoid the situation where technology determines everything. We should carefully
analyze the blockchain technology system and the neoliberalism behind the tech-
nology and perform a detailed evaluation of how the rule of code will affect social
life.

5.3 Applications of Blockchain in Rule of Law

With the rapid development of blockchain technology, “blockchain technology+”

justice has proliferated the rule of law and has bright prospects. Since the application
of blockchain in the rule of law is a new field, very little news can be found about
it in foreign jurisdictions. In China, the birth of the Internet Court has accelerated

121 See [FR] Primavera de Filippi, [US] Aaron Wright, Regulating Blockchain: the Rule of Code,
translated by Wei Dongliang, CITIC Press, 2018 edition, pp. 84–89.
122 See Gidong Chen, “Sowing the Technical Ghost of Trust - A Review of Blockchain Legal

Research,” Exploration and Controversy, Vol. 12, 2019.

192 Y. Cui

blockchain technology application in the rule of law. China has led the world in
the application of blockchain technology in the rule of law. China’s “blockchain
technology+” rule of law is mainly used in building judicial depository and anti-
tampering and anti-forgery platforms.

5.3.1 Overview of Application of Blockchain in Rule of Law

Cities in East China Took the Lead in Blockchain Application in Rule of Law

Cities in East China have more favorable conditions for the application of blockchain
in the rule of law, with clusters of blockchain talent, advanced technology and
strong financial support. The judicial departments in first-tier cities, such as Beijing,
Shanghai and Guangzhou, broke the mold in exploring the application of blockchain
technology in rule of law to establish advantages in both ideology and technology
fields as well as technology applications. Meanwhile, local governments in these
cities also encourage innovation and provide financial support to judicial departments
and soon form a leading edge in the field.

Cities in Central and Western China Established Distinguished Platforms

for Blockchain Application in Rule of Law

Blockchain application in the rule of law in central and western cities in China largely
depends on local judicial departments since these regions have not yet laid a solid
foundation in the field. Judicial departments with innovative ideas and capabilities
can catch up and form their own characteristics. Zhengzhou Intermediate People’s
Court, Hefei Shushan District People’s Court and Pidu District People’s Court in
Chengdu have built their own blockchain deposition platforms.

5.3.2 Features of the Application of Blockchain in Rule of Law

Uneven Development of the Application of Blockchain in the Rule of Law

(1) The imbalance in the field of legal application of blockchain technology. From
the perspective of the application field, the application in the judicial field
is increasingly advanced. The application in the law enforcement field is in
the exploratory stage, and its application in the legislation field has not been
reported. For the application injudicial field, the courts have the most typical
cases.
(2) The imbalanced legal application area of blockchain technology. Due to dispari-
ties in talents, technologies, ideas and financial resources, cities in eastern China
took the lead in blockchain application in rule of law, and cities in central and
western China established distinguished platforms.
5 Special Reports on Development of Artificial Intelligence … 193

Blockchain Application in Rule of Law Largely Depends on Financial

Support from Governments

Governmental financial support is a major driver of blockchain application in the

rule of law. Currently, there are no commercial products for blockchain application
in the rule of law, and all products need to be customized. For that reason, generally,
Party A will be responsible for product development and apply for a budget from the
government, then purchase the products by means of government procurement. Data
from Interchain Pulse, a blockchain information platform, showed that there were 14
government tenders related to blockchain in China in 2018, and in 2019, the number
was 31. Among them, most bids were from Beijing, while local governments such
as Guangdong, Shandong, Shanghai and Gansu also participated.

Key Technology Remains a Weak Link in Blockchain Application in Rule

of Law

Major developed countries in the world invested in the research and development
of the key core technology of blockchain earlier. After making pioneering break-
throughs, they mainly used it in the field of cryptocurrency, leading the era of
blockchain technology 1.0 and 2.0. With blockchain 3.0 approaching, China has
carried out much application-level research and innovation with the help of open-
source core technology and has made significant achievements in specific appli-
cations, especially in the judicial field. Attention must be paid to developing core
technology since it provides core competencies.

The Development of Blockchain Applications in Rule of Law Calls

for Interdisciplinary and Cross-Sectoral Work

The government needs to focus on supporting the research, development and appli-
cation of key technologies such as blockchain storage, encryption, consensus and
cross-chain technology, at the same time, promote both interdisciplinary and cross-
disciplinary research related to blockchain technology, make the best of resources
in government, higher education institutions, companies and industry organizations,
cultivating a favorable environment for the innovation of blockchain technology.

5.3.3 Future of Blockchain Application in Rule of Law

Implementing the National Plan and Developing Core Technology

The further application of blockchain technology in rule of law calls for national
plans and a good command of key technology. President Xi stressed that we should
strengthen basic research ability and boost innovation capacity to lead theoretical
194 Y. Cui

research, innovation and industries of the emerging field. We should advance coordi-
nated key tasks tackling and accelerate breakthroughs in key technologies to provide
safe and controllable technological support for blockchain development and its appli-
cation. We should step up research on the standardization of blockchain to increase
China’s influence and rule-making power in the global arena. We should promote
industry development, make the best of market advantage, promote blockchain inno-
vation and application. We should deepen the integration between blockchain and
other information technologies, including artificial intelligence, big data and the
Internet of Things. We should establish a complete talent cultivation system and
nurture quality innovation team building.

Improving Policies and Regulations and Providing Targeted Management

The rule-of-law application of blockchain technology relies on efficient policies and

regulations. Guidance and regulations are needed in blockchain technology devel-
opment. We must analyze potential risks in the field, keep track of its trend, and
meaningfully explore new possibilities. We must establish a security system that fits
blockchain technology architecture that regulates blockchain developers and trans-
action platform operators and ensure that they act in accordance with the law. We
should introduce the idea of rule of law in blockchain network management and
ensure its safe and healthy development.

Promoting Both Blockchain Technology and Related Policies and Regulations

The integration of blockchain technology and rule of law calls for the collective
efforts of experts from both legal and technological fields. The future of blockchain
technology, whether keeping the technology on a tight rein or giving free rein to
it, relies on how policies and laws can regulate its development. As with any other
technologies, the development of blockchain technology starts with breaking current
regulations and will eventually adapt to new regulations as it evolves. In accordance
with the general law of scientific and technological development, the development
of blockchain technology depends on whether there is an effective and legal gover-
nance structure, which requires codes that can control the operation of digital tech-
nology and, more importantly, policies and regulations to regulate it. It is a historical
necessity that blockchain technology and policies and regulations promote each other.

5.4 Analysis of Blockchain-Related Cases

In recent years, blockchain technology has proliferated people’s social and economic
lives, and the number of blockchain-related cases is on the rise. This section provides
an in-depth analysis of these cases to generate solutions for legislation.
5 Special Reports on Development of Artificial Intelligence … 195

In this part, we conducted an in-depth investigation of blockchain technology and

judicial cases. We chose blockchain as a keyword for domestic case investigations on
China Judgments Online (https://wenshu.court.gov.cn/). As of December 31, 2019,
we analyzed 142 copies of judgments on their locations, types of technologies and
causes, and the findings are as follows.

5.4.1 Regions

The development of blockchain technology varies in different regions. This part

analyzes the regions where the adjudication courts are located in 142 cases and
determines the regional differences.

Blockchain-Related Cases Were Concentrated in Eastern China

Most blockchain-related cases occur in eastern China, such as Beijing, Zhejiang, and
Guangdong, which account for 63% of the total case volume. These regions have
a better economic environment that attracts many talents and resources in the field,
and as a result, the number of blockchain-related disputes is on the rise.

Blockchain-Related Cases Spreading to Central and Western China

Blockchain-related cases were also found in Henan, Hunan, Hubei, Sichuan and
Xinjiang. Therefore, we should plan ahead and actively draw experience from these
cases so that the law can better serve the economy and society.

5.4.2 Analysis of the Type of Technology Involved in the Cases

The application of blockchain technology includes data, smart contracts and feder-
ated blockchain. This section provides an in-depth analysis of these technologies to
provide a reference for subsequent case adjudication.

Cases Involving Data Technology Accounting for More Than Half

Data-related cases account for half of blockchain-related cases. The application

of blockchain technology revolves around data, which has become an important
strategic resource, and the country should attach more importance to data protection.
196 Y. Cui

The Cases Involving Smart Contract Technology Accounting for the Second
Most

The rise of smart contracts, a hot topic in legal research, is a major breakthrough
in the Blockchain Technology 2.0 era. A smart contract codifies a specific contract.
It is a set of digitally defined commitments written into computer-readable code
that can be executed by a computer or network of computers. In most cases, smart
contracts are indirectly used. On December 19, 2019, the Hangzhou Internet Court
online pronounced the first case in China in which the smart contract technology of
judicial blockchain was directly applied to the trial procedure of civil litigation, and
it was also the first case in the field of e-commerce in which the whole process of
transaction was stored through blockchain.

Cases Involving Federated Blockchain Accounting for the Third Most

As an important type of blockchain technology, federated blockchain is also used in

some cases. In the copyright infringement cases between Chineseall Co., Ltd. and
Guangzhou Dongjing Computer Technology Co., Ltd., the court determined that
the defendant operated an application that provided an online listening service for
the alleged works because the digital evidence presented by the plaintiff is reliable
in its production, storage and content integration, and its authentication should be
confirmed since no counterevidence was provided and should be taken as primary
evidence.

5.4.3 Causes of Cases

This part analyzes four major causes of cases, among which the causes of civil cases,
enforcement cases, criminal cases and administrative cases were 112, 16, 13 and 1,
respectively.

Most Causes of Cases Were Civil Cases

Seventy-nine percent of cases were civil cases, most of which were civil disputes,
especially disputes over intelligent property rights, cooperation agreements and
contracts. Therefore, more efforts should be made to protect blockchain technology
intelligent property, encourage and protect innovation and strive to make break-
throughs in key core technologies. Otherwise, the industry is not willing to increase
investment in R&D, and the development of the whole industry will be seriously
restricted.
5 Special Reports on Development of Artificial Intelligence … 197

Causes of Criminal Cases Were on the Rise

The number of causes of criminal cases was 13, accounting for 9% of the total,
including 1 in 2017, 4 in 2018 and 8 in 2019. The exponential increase year by
year should receive great attention. Criminal cases involving blockchain technology
mainly include illegal absorption of public deposits, fundraising fraud, organization
and leadership of pyramid schemes, which involve large groups of victims and are
very likely to trigger large-scale rights protection activities and mass gatherings.
Therefore, we must attach great importance to it and try our best to ensure the safety
of the legitimate property of the public.

6 Empirical Research on Training Interdisciplinary Talents

for AI and Rule of Law123

6.1 Foreign Theoretical Research and Practice of AI

and Rule of Law

Since “artificial intelligence” was first introduced at Dartmouth in 1956, it has been
developed for more than 60 years and widely applied to politics, economics, diplo-
macy, education, health and other fields. The rapid development of AI inevitably
brings with it the need for institutions, rules, standards and procedures, which is
naturally closely integrated with the law. In 1970, a law review with the title Some
Speculation about Artificial Intelligence and Legal Reasoning published by Bruce
Buchanan and Thomas E. Headrick indicated that the interdisciplinary studies of
“AI and Law” have become one of the branches of AI. Since the early 1970s, Ward
Pop and Benhard Schlinker have developed a computer program to advise lawyers
in reasoning cases called “JUDITH”, and D. Waterman and M. Piteson developed a
legal judgment aid system (LDS) in 1981. Since the late 1980s, American AI experts
have begun to pay attention to interdisciplinary studies of AI and law. The biennial
International Conference on Artificial Intelligence and Law (ICAIL) with its subcon-
ferences and magazines, such as Artificial Intelligence and Law, have been important
academic platforms for researchers to determine the future development of AI and
Law.
In general, the features of AI research aboard are mainly demonstrated in the
following aspects. First, the interdisciplinary approach has always been adopted
by foreign researchers, as several papers given at the World Robot Conference are
jointly published by scholars in the fields of law, engineering and philosophy. Second,
research is featured with the joint effort made by both theorists and practitioners.

123Introduction to the Authors: Liu Xiaohong, President of Shanghai University of Political Science
and Law, Professor, Ph.D Mentor. Yang Hua, Dean of the Law School of Artificial Intelligence
in Shanghai University of Political Science and Law, Professor, Ph.D Mentors. Niu Bingru,
Postgraduate of Shanghai University of Political Science and Law.
198 Y. Cui

Specifically, the academic community is responsible for the theoretical research, and
industry and government are mainly in charge of technological development and
policy guidance, respectively. Each participant plays its own advantages actively to
explore and regulate AI, making the integration of industry, politics and academia
a a norm. Third, most of the theoretical research results are based on the practical
application of AI, discovering problems and reflecting on social impacts in practice,
and theoretical research is close to the practical needs of society. Fourth, there are
various theoretical research perspectives, both horizontal and longitudinal research.
Scholars believe that the management and control of AI should vertically refer to the
coping strategies of the industrial revolution, both of which are social changes driven
by technology. Horizontally, scholars integrate the governance practices of various
countries and explore localization research in their own countries. Finally, smooth
progress has been seen in policy guidance and legislation. In terms of policy support
and the formulation of ethics laws and regulations, many legislative attempts have
been made, such as directly amending laws or issuing new laws. For example, in the
field of autonomous driving, countries such as the United States and Germany grant
AI legal status in experiments and road tests to encourage its industrial application.
At the same time, the state funds universities and research institutions to conduct
theoretical research and establish a policy system to support the development of AI.
At the same time, government-funded universities and scientific research institutions
are engaged in fundamental research that promotes the establishment of a policy
system to facilitate the development of AI.
The rapid development of AI will profoundly change human society and the world
as a whole. Therefore, no country wants to lag behind in the game of AI. For the
United States, to maintain its leadership in AI, the White House released two impor-
tant reports called the National Artificial Intelligence Research and Development
Strategic Plan and Preparing for the Future of Artificial Intelligence in October
2016. As the world’s first national-level strategic plan for AI development named the
new “Apollo Project” by Barack Obama, it mentioned seven recommendations for
AI strategic development with goals of research investments, which includes devel-
oping AI collaboration methods, underscoring the need to address the ethical, legal,
and societal implications of AI, emphasizing safety and security, and developing
the standards and benchmarks for the evaluation of AI technologies. Preparing for
the Future of Artificial Intelligence was issued to prepare and guarantee AI develop-
ment in terms of policy formulation, government supervision of technology, financial
support, education of artificial intelligence for the public, prevention of machine bias,
etc., and puts forward twenty-three suggestions for the implementation of AI. In June
2019, the White House published an update of the National Artificial Intelligence
Research and Development Strategic Plan, and a new recommendation “focusing on
the increasing importance of effective partnerships between the federal government
and academia, industry, other nonfederal entities, and international allies to generate
technological breakthroughs in AI” was put forward.
Because AI will challenge the moral baseline of human beings and even pose
unpredictable risks, it is necessary to enact and improve related laws, regulations,
5 Special Reports on Development of Artificial Intelligence … 199

systems and ethical and moral systems to forge a sound environment for AI develop-
ment. At the same time, AI and law, an interdisciplinary study that focuses on legal
phenomena and rules, have come into being, leading and promoting AI development
with certain regulations and guarantees.
In terms of higher education and research institutions, U.S. law schools have
offered many courses of “AI and Law” since the 1980s. Stanford Law School is the
pioneer in “AI and Law” studies. Currently, Stanford Law School has set up the Stan-
ford Center for Legal Informatics (CODEX) to bring scholars, lawyers, entrepreneurs
and technology professionals together to explore the future of legal automation. In
addition, Stanford has also set up a profound program with the title of “Law, Science
and Technology” for the Master of Laws, providing them with systematic and exten-
sive courses covering law and technology. Since the beginning of 1985, Harvard Law
School has arranged “AI and Law” seminars followed by Stanford University. In the
spring of 2017, Harvard Law School and the Massachusetts Institute of Technology
(MIT) Media Lab collaborated to host an innovative January-term course called
“Internet & Society: The Technologies and Politics of Control.” Moreover, a new
teaching program called the Harvard Law School Legal Clinic was also initiated by
Harvard. Another American law school, Northeastern University School of Law with
a long history of teaching and research in “AI and Law”, offered seminars in 1987. At
present, the centers of law and innovation are mainly responsible for the research for
“AI and Law”. Georgetown Law School, in addition to offering courses concerning
this field, organizes an annual Georgetown Law’s Iron Tech Lawyer Invitation to
encourage students to apply new technologies to legal services with innovation.
A number of innovative interdisciplinary courses for “AI and Law” are offered to
students in most law schools of American public universities, such as Michigan State
University, University of Pittsburgh, University of Minnesota System, University of
Washington, Chicago-Kent College of Law and Boston Suffolk University.
An increasing number of law schools are offering courses concerning AI around
the world. For example, the University of Copenhagen in Denmark has opened a
course named “Robot Law: Artificial Intelligence, Robot Technology and Law”.
The University of Auckland Law School in New Zealand has opened “Artificial
Intelligence: Law and Policy”, and the University of Edinburgh Law School has
introduced a major called “Robotics and the Law”.
In addition, most well-known universities’ law schools have set up research
institutes of “AI and Law”, which include the Stanford Center for Legal Infor-
matics (CODEX), the MIT Computational Law Laboratory, the Center for Innovative
Law and Policy Studies at the University of Toronto, and the Center for Artificial
Intelligence and Law at the National University of Singapore.
200 Y. Cui

6.2 Theoretical Research and Practice of Artificial

Intelligence and Law in China

China is a latecomer to explore the interdisciplinary study of AI and law, and its
research was relatively immature approximately 22,000. Since 2017, legal profes-
sionals have realized that the legal issues brought by AI should be taken seriously.
However, substantial progress has not yet been seen due to the complexity of AI
governance, the lag of its legal system development and other factors. The devel-
opment of the AI industry has not yet received much attention due to the lack of
exchanges with other disciplines. Most research in our country focuses on academic
and theoretical outcomes instead of key areas and practical AI applications. In partic-
ular, many scholars have begun to be concerned about the educational reform of AI
and law. On October 21, 2018, China Higher Education and Southwest University of
Political Science and Law cohosted the First Artificial Intelligence Law Education
Forum, providing opportunities for participants to exchange and discuss “AI and
Law”, such as the dramatic influence and reform brought by the legal education in
AI world as well as the impact on the legal profession with its response strategy.
In China, judicial practice concerning AI and law has taken the lead over theoret-
ical studies. Since 2002, the Supreme People’s Court has issued a series of guiding
documents for informatization. On January 29, 2016, the Supreme People’s Court’s
Leading Group on Information Construction held its first plenary meeting in 2016,
proposing the frontier “smart court” and releasing The Construction Planning of the
Information 3.0 Of the People’s Court. In July 2016, the General Office of CPC
and the State Council released the Outline of National Informatization Develop-
ment Strategy, which pointed out that the construction of “Smart Courts” should
be put on the agenda. The Supreme People’s Court issued the Wuzhen Consensus,
expressing the determination to build “Smart Courts”. On December 15, 2016, the
State Council issued the 13th Five-Year Plan for National Informatization, which
clearly states its support to the “Smart Courts” and the promotion of electronic liti-
gation and judicial informatization. On April 20, 2017, the Supreme People’s’s Court
issued the Opinions of the Supreme People’s Court on Accelerating the Development
of Smart Courts. Meanwhile, local courts also actively promoted the application of
AI in proceedings and developed various AI-driven robotic legal systems, such as
the AI-powered judges (Rui Faguan) claimed to be “the first of its kind in the world”
launched by the Beijing Internet Court, the “206” Intelligent Auxiliary Case Handling
System adopted by the Shanghai Courts and “the Smart Trail System” in Suzhou. In
April 2018, the Supreme People’s Court released the Smart Court Evaluation Report
(2017), revealing that the fundamental support capacity index is the highest among
the “three major functions” of the national court with network informationization,
which is up to 90. This indicates that the full coverage of information infrastruc-
ture was almost achieved by the “Integration of Cloud Computing and Internet”.
On March 1, 2019, the Annual Report on Informatization of Chinese Courts (Blue
Book of Rule of Law) was released to show that the construction of smart courts
5 Special Reports on Development of Artificial Intelligence … 201

featuring networking, transparency and intelligence has seen significant progress

with comprehensive functions.
For the effort made by higher education and research institutes, Tsinghua Univer-
sity has set up an interdisciplinary J.M. program in Computational Law with a
series of technical courses concerning the Internet, Big Data and AI. In addition,
the Law School of Renmin University of China has also offered interdisciplinary
courses, which reform the teaching of “Law + Science and Technology”. For Sichuan
University, a Legal Data Analytics course was offered to Juris Doctor.
Currently, domestic research institutes for “AI and Law” include: Working Group
on polices and regulations of China’s Artificial Intelligence Industry Alliance (AIIA),
The Institute for Studies on Artificial Intelligence & Law of Tsinghua University, the
Legal Artificial Intelligence Laboratory of Peking University, the Legal and Artificial
Intelligence Research Center of Peking University, Law and Technology Institute and
Smart Procuratorate Innovative Institute of Renmin University of China, Big Data
and Artificial Intelligence Research Center of China University of Political Science
and Law, Institute of Logic and Cognition of Sun Yat-sen University, Big Data Lab
of Sichuan University, the Institute for Studies on China’s Artificial Intelligence and
the Rule of Law of Tiantifin University, Law at AI Lab of Soochow University, the
Institute for Studies on AI and Law Justice Center of Chongqing Collabor Intelligence
and the Rule of Law of Tiantificial University of AI and Law Law University of
Soochow University of Political Science & Science. On December 6th, the first Law
School of Artificial Intelligence of Southwest University of Political Science and Law
was established and began to enroll students in 2018. In May 2019, the Law School
of Artificial Intelligence established by Shanghai University of Political Science and
Law accepted 40 students as the first batch of undergraduates to learn the major “AI
and Law”.
Conferences with a specific focus on “AI and Law” includes: the Rule of Law
Forum of World Artificial Intelligence Conference and the Seminar on “Legal Artifi-
cial Intelligence and Social Governance” jointly sponsored by Shanghai Law Society
in 2018 and 2019, Youth Rule of Law Forum cohosted by Shanghai Law Society and
Shanghai University of Political Science and Law in 2019, the first China “AI and
Law” Challenge (CAIL 2018) and its seminar held by the Supreme People’s Court
Information Center, the first Artificial Intelligence and the Rule of Law Forum hosted
by Zhejiang University Guanghua Law School (2018), the first Annual Forum On
Artificial Intelligence and Legal Education cohosted by the magazine named China
Higher Education and Southeast University of Political Science and Law (2018),
Tsinghua Computational Law Forum (2018) and its summer camp (2018–2019),
Spring School on Argumentation in AI & Law (SAIL-2018) established by the Insti-
tute of Logic and Cognition of Sun Yat-sen University, the Empirical Legal Summer
Camp held by Sichuan University and so on.
202 Y. Cui

6.3 The Promotion of the Research on AI and the Rule

of Law by Shanghai Law Society

During the World Artificial Intelligence Conference (WAIC) held in Shanghai on

September 17th to 19th in 2018, Shanghai Law Society and other related organi-
zations cohosted the AI and Law Forum, which is one of the four special events
mentioned in WAIC 2018. The Shanghai Initiative of Artificial Intelligence and the
Future Rule of Law was issued in the conference with 14 practical recommendations
in four aspects.
To promote the study of AI and the Rule of Law, the Shanghai Law Society and
the Yangtze River Delta Research Institute of Zhejiang Province jointly published
the Blue Book on World Artificial Intelligence Rule of Law (2019), which compre-
hensively demonstrates the theoretical research and practical progress of AI and the
Rule of Law at home and abroad. On August 30, 2019, the Rule of Law Forum
of WAIC 2019 cohosted by the Organizing Committee of WAIC, Shanghai Law
Society and China Justice Big Data Institute was held in Shanghai Exhibition Center
accompanied by the “Writing Essay Activity” and the Blue Book on World Artificial
Intelligence Rule of Law (2019). Moreover, The Guidance on Artificial Intelligence
Security and the Rule of Law (2019) was jointly released in the Rule of Law Forum
of the WAIC and the High-Level Dialog on World Artificial Intelligence in 2019.
The Youth Rule of Law Forum, a specific session of the WAIC 2019 toward youth
around the world, released the initiative called “Youth Responsibility in the Age of
Artificial Intelligence”. Representatives of universities in the forum also signed the
“Artificial Intelligence and the Rule of Law Alliance Agreement.“
To advance the national strategy of AI, the Party Group of Shanghai Law Society
approves the establishment of its Artificial Intelligence and the Rule of Law Research
Association with deliberation, which is not only a trail-blaze move but also a concrete
action to build the “Artificial Intelligence Ecosystem” in Shanghai.

6.4 The Establishment and Development of the School

of Artificial Intelligence and Law, Shanghai University
of Political Science and Law

In 2017, the Notice of the State Council on Printing a New Generation of Arti-
ficial Intelligence Development Plan was enacted to implement the plan of early
AI development made by the CPC Central Committee and State Council to seize
great strategic opportunities in the AI-driven world and develop China’s first-mover
advantages. General Secretary Xi Jinping provided important instructions for AI
group studies in 2018, stressing that universities should strengthen AI professional
training and improve the design of AI-related courses and majors. In April 2018,
another document called the AI Innovation Action Plan for College and Universities
was introduced to publicize the idea of “AI Plus X” (“X” hereinafter referred to as
5 Special Reports on Development of Artificial Intelligence … 203

another discipline except for AI). In 2019, the Ministry of Science and Technology
issued The Notice on the Publication of the Guidelines for National New Gener-
ation Artificial Intelligence Innovation and Development Pilot Zone Construction
Work. On January 21, 2020, the three ministries, Ministry of Education, the National
Development and Reform Commission, and the Ministry of Finance, jointly issued
Opinions on Promoting the Integration of Disciplines and Improving the Training
of Postgraduates in the Field of Artificial Intelligence with the Lead of “Double
First-class” Universities (“the Opinions”). The Opinions points out that China is
now catching up with other countries in AI development. Therefore, it is necessary
to advance interdisciplinary study by strengthening the training of professionals in
specific fields, cultivating more talent in basic theoretical studies, building a training
system in which the basic theoretical talent and composite talent of “artificial intel-
ligence + X” are treated on an equal basis, and exploring ways to promote the deep
integration of AI and other disciplines and the training of its talent. It can be seen
that the CPC Central Committee, the State Council and state ministries attach great
importance to the discipline, industry and personnel training concerning AI, which
lays a foundation for the development of the major “AI and Law”.
Comrade Li Qiang, Secretary of the Shanghai Municipal Party Committee, said
at the opening ceremony of the 2018 World Artificial Intelligence Conference that
Shanghai will focus on “Innovative Sources”, “Application Demonstrations”, “Insti-
tutional Supply” and “Talent Pooling” to accelerate the construction of the AI Hub
in Shanghai. Among the “four hubs”, “institutional supply” and “talent pooling” are
closely related to the artificial intelligence law.
In accordance with the major project called the “China National Institute for SCO
International Exchange and Judicial Cooperation” instructed by General Secretary
Xi Jinping, Shanghai School of Political Science and Law attaches great importance
to research on national security by enacting AI-related laws and stressing the ethics
of AI utilization. To execute the tasks of “Institutional Supply” and “Talent Pooling”
proposed by the Shanghai Municipal Committee, the first School of Artificial Intelli-
gence and Law was established in Shanghai with the support of Shanghai Law Society
and IFLYTEK CO., LTD. in May 2019, enrolling 40 undergraduates in the major of
“AI and Law”. Currently, it has initially formed a specific mode for interdisciplinary
research and talent cultivation. To promote the study of AI and the Rule of Law, the
school is involved in preparing the Blue Book on World Artificial Intelligence Rule of
Law (2019). Since August 30, 2019, supported by Shanghai Law Society, Shanghai
Municipal Commission of Economic and Informatization and other organizations,
the annual “Youth Rule of Law Forum, World Artificial Intelligence” has held with
the purpose of building an international platform for young people around the world
to express their ideas, enhance their capacity and assume responsibility in the AI
world. In addition, as the secretariat unit of the Artificial Intelligence and the Rule
of Law Research Association of the Shanghai Law Society, the school also makes
great efforts to provide a full range of support and services for researchers of AI and
the Rule of Law in Shanghai and even the whole country. The talent training mode
of AI and law was admitted to the talent development programme of the Shanghai
Municipal Education Commission. In addition, the construction of Shanghai Think
204 Y. Cui

Tank requires more systems, rules, standards and procedures. For higher education,
courses such as “Artificial Intelligence and Law” and “Programme Design for Data
Analysis” are admitted to the Shanghai University Curriculum Teaching Reform
Project.

6.5 The Future Development of the School of Artificial

Intelligence and Law of Shanghai University of Political
Science and Law

The blueprint of the Law School of Artificial Intelligence and key points of work were
formulated at the opening of the Law School of Artificial Intelligence (the School).
On September 23, 2019, a meeting was convened on the establishment of the School,
and The Minutes of the Law School of AI Work Coordination Meeting (The Minutes)
was issued. The School has formulated The Construction Plan of the Law School
of Artificial Intelligence 2020–2022 to further implement the instructions given by
general secretary Xi Jinping, the Opinions jointly issued by three ministries and the
specific contents of the Minutes.

6.5.1 Development Orientation

To advance the national AI strategy and cultivate interdisciplinary talent to practice

AI and law, two education plans are mapped out to meet the goal. One is to culti-
vate interdisciplinary talents who are able to handle the legal issues arose from AI
application, including those who provide legal services for the four legal aspects of
artificial intelligence, such as legislation, law enforcement, judicial and law-abiding.
The other is to cultivate more practical talent to improve the application of AI in the
four legal aspects mentioned above.

6.5.2 Talent Cultivation Programme

Talent cultivation, including both undergraduate talent cultivation and postgraduate

talent cultivation, is one of the fundamental tasks of the university.

The Tasks and Objectives of the Undergraduate Programme

The talent development orientation of Shanghai University of Political Science and

Law is to “take undergraduate education as the fundamental”. The three-year plan for
undergraduates in the school includes the goals of improving the design of curriculum
in undergraduate teaching, carrying out the construction of the second phase of the
5 Special Reports on Development of Artificial Intelligence … 205

Comprehensive Laboratory of “AI and Law”, providing internship opportunities for

students in courts, procuratorates and public security systems, law firms, and well-
known enterprises of AI, developing systems of practical experiences that serve as
teaching materials, adding undergraduate majors beyond the curriculum of law, such
as “Artificial Intelligence and Jurisprudence” and “Practical Legal Study of Artificial
Intelligence”, to comprehensively support the development of the school.

The Tasks of the Postgraduate Programme

While the school focuses on the education of undergraduates, the LL.M. concerning
AI should be taken seriously as well. In addition, the curriculum and the training
programs for the Ph.D. of “AI and Law” should also be developed actively when it
is well prepared.

6.5.3 Features

To develop the major “AI and Law” by applying related theories to practice, empha-
sizing the systemic combination of “AI and Law” and focusing on improving the
practical capacity of students. The school takes advantage of the policies, industry and
technical preponderance of substantive offices and enterprises and works closely with
legal practical departments to actively explore the establishment of school-enterprise
cooperation, integration of production and education, combination of learning and
research, and cross-integration of artificial intelligence and law. For the curriculum,
the courses should be featured with interdisciplinarity, practicality and applicability.
In addition to the core legal courses prescribed by the Ministry of Education, some
courses covering the frontiers of artificial intelligence technology and some inter-
disciplinary courses combining artificial intelligence with law will be offered to
students.

6.5.4 The Development of Discipline

In the future, the school will develop “AI and Law” based on advanced research
findings, adopt “full-time” and “flexible” ways to introduce talent through multiple
channels, improve the training of teachers and hold high-level academic exchanges
with platforms, seminars and academic publications such as the Youth Rule of Law
Forum in World Artificial Intelligence, China Marine Strategy Forum, the Marine
Law Research Association and Artificial Intelligence and the Rule of Law Research
Association of the Shanghai Law Society. Selected teachers and students will be
sent abroad for further study, academic exchange and work, aiming to broaden their
horizons and enhance their teaching, scientific research and practical capability. With
the help of research institutes such as the Justice Research Association and the Rule
206 Y. Cui

of Law Research Association, the school will contribute to improving the decision-
making quality. Think tanks are given priority in providing advice to the state and
local governments to serve society in a better way.
Chapter 6
The Use of AI in Law in Response
to Major Epidemics

Yadong Cui

1 “Law + AI” Empowering Governance During

the COVID-19 Epidemic1

When the state was pursuing modernized national governance, the unexpected
COVID-19 epidemic became a test for the system and the competence of national
governance. The essence and core of modernized national governance is the rule of
law, and the governance of epidemics requires the rule of law. At the third session
of the Rule of Law Committee of the Central Committee of the Communist Party
of China on February 5, 2020, General Secretary Xi Jinping emphasized that we
should use the thinking and method of the rule of law in the control of epidemic and
improve the competence of the legal control and the governance of epidemic in every
links including legislation, law enforcement, justice and law abidance to support the
governance of epidemic with the rule of law. Epidemic control in accordance with
laws indeed plays a crucial role in legally advancing epidemic control and ensuring
the smooth governance of epidemics.
As the control of the epidemic started, General Secretary Xi Jinping specifi-
cally required science-based approaches to the control of epidemic and targeted
measures at a meeting of the Standing Committee of the Political Bureau of the
Central Committee of the Communist Party of China on February 12. On March 2,
General Secretary Xi Jinping highlighted the usage of AI, big data and other new
technologies to provide scientific and technological support in the war against the
virus when inspecting research on COVID-19.
Thus, both law and AI are essential for the governance of epidemics.

1Author: Yang Hua, dean, professor and Ph.D. supervisor of Artificial Intelligence and Law School,
Shanghai University of Political Science and Law.

Y. Cui (B)
Shanghai Law Society, Shanghai, China
e-mail: [email protected]

© Shanghai People’s Publishing House 2022 207

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2_6
208 Y. Cui

1.1 The Importance of Law in Epidemic Control

Deficiencies in the alert and response plans for emergencies such as epidemics
were exposed not only in the prevention of the COVID-19 epidemic but also in
the following control of the epidemic. The epidemic came unexpectedly because it
was the first time that COVID-19 was identified by humans, situations in different
regions varied, and the relevant competent departments adopted diverse coercive
measures. In China, the measures preliminarily involved civil, administrative and
even criminal factors. If they were mishandled, both the vital interests of people and
the rule of law in the state would be compromised.
In the governance of the epidemic, those prosecuted for endangering public secu-
rity, illegal production and selling fake and inferior products or other crimes included
not only ordinary people, medical supplies manufacturing companies and their own
legal representatives but also staff who perform the duty of epidemic prevention and
control. Those who hurt medical personnel with violence or committed disturbing
acts were prosecuted for the crime of causing disturbance in accordance with laws.
Specific measures of the rule of law were embodied by regulating donation and accep-
tance, strengthening public security administration, maintaining market regulation,
regulating public opinions and publicity and improving the administration of duty.
At present, work resumption and the movement of people have begun. However,
the more critical the epidemic is, the more emphasis should be placed on control in
accordance with laws.
It is of practical significance for promoting the construction of government and
society under the rule of law to use the thinking and way of rule of law to deal with
epidemic crisis and deal with major public health emergencies Led by the “Four
Pronged” Strategy, the epidemic control in accordance with laws has superiorities
for the following reasons:
First, epidemic control in accordance with laws can guarantee authorized, consis-
tent and efficient epidemic control policies. Law is backed by the coercive power
of the state. Epidemic control in accordance with laws can guarantee the effective
implementation of epidemic governance policies and measures of the Party and the
State and follow the principles of epidemic governance of the Party and the State.
During epidemic control, the Party and the State adopted the philosophies of the rule
of law and epidemic control in accordance with laws, and governments in different
regions published respective public health emergency response levels in light of the
local situation and in accordance with the Law of the People’s Republic of China
on Prevention and Treatment of Infectious Diseases and the Law of the People’s
Republic of China on Emergency Response. In addition, both the central government
and local governments have enacted a series of epidemic control policies and regula-
tions, such as the Notice of the Joint Prevention and Control Mechanism in Response
to the COVID-19 Epidemic of the State Council on Prevention and Control Measures
for Key Locations, Organizations and Population Groups. Thus, by means of legal
documents, the effect of epidemic prevention is guaranteed, and epidemic control is
supported by laws and regulations.
6 The Use of AI in Law in Response to Major Epidemics 209

Second, epidemic control in accordance with laws can guarantee a stable, ongoing
and sustainable society. At the outbreak of the epidemic, people were on edge, and
at the beginning of the epidemic, the lack of effective medicine and treatment caused
panic in society. In the practical field of epidemic control, there were different chaos at
the beginning, such as village closure, road destruction, malicious price gouging, and
transportation blocking of epidemic prevention supplies. However, as relevant laws
and regulations were enacted and law enforcement departments strictly complied
with laws during duty performance and case solution, this chaos gradually disap-
peared. Preliminary results of the promotion of a rule of law society were achieved.
The governance of epidemics requires diverse policies to protect multiple inter-
ests. Only by implementing these policies in accordance with laws can we reach a
consensus and guarantee orderly epidemic control in society.
Third, epidemic control in accordance with laws can guarantee transparent, oper-
able and expectable epidemic control policies. A legal system is the transparent rule
of rules and procedures. The public can act in accordance with laws and predict the
consequence. During the epidemic, various cases related to the epidemic emerged.
However, no matter what kind of cases they are, if they are about epidemic control,
the life, health and safety of people are priorities, and the way to effectively protect
them is the rule of law. We should take strict control and measures against speculation
and hoarding of epidemic control supplies for profits during the epidemic and focus
on the effectiveness and operability of epidemic control policies so that laws produce
good normative effects. Every little epidemic control is closely related to the public,
and the normative procedure of policy implementation should be promised so that
the public and companies can have expectations about epidemic control.
Fourth, epidemic control in accordance with laws can make epidemic control
policies more effective, sanctionable and remediable. The coercive power of law
backed by the state guarantees effective epidemic control polices. Therefore, any
organization violating laws and regulations will be sanctioned. Anyone whose inter-
ests are jeopardized can seek remedies in accordance with laws. During epidemic
control, the Supreme People’s Court and the Supreme People’s Procuratorate have
published a series of guiding cases related to the epidemic, which effectively provide
local courts with trial models when hearing similar cases, and by the evaluation of
people’s behaviors, they guide the public to act orderly.
An epidemic tests not only the system and the competence of national governance
but also the competence and the level of the rule of law. In epidemic governance, there
are dozens of relevant criminal crimes. In addition, there are crimes of obstruction
of infectious disease control, illegal trapping and killing precious and endangered
wildlife, illegal hunting, production and sale of foods that do not meet safety stan-
dards, production and sale of toxic and hazardous foods, obstruction of public affairs,
graft, abuse of power, negligence, embezzlement and so on.
In epidemic control in accordance with laws, criminal punishment is the most
severe sanction. If relevant provisions of other laws, such as administrative law, civil
law, commercial law and economic law, are also considered, epidemic control in
accordance with laws can be basically provided with a multidimensional legal system.
Certainly, some laws and regulations on the utilization of high technology should
210 Y. Cui

be further improved. Therefore, it is necessary to further improve the legislation

related to epidemic governance, step up the establishment of supporting systems,
perfect punishment procedures, strengthen public security and build a legal system
for epidemic governance that is complete, science-based, normative and effective.
Everything is a double-edged sword. An epidemic is a crisis, but a response is an
improvement. Upon the response to the epidemic crisis, the governance competence
of the Party and State will be further improved, the legal awareness of people will
be further heightened and the rule of law competence will be further enhanced.

1.2 The Importance of AI in Epidemic Control

Upon the outbreak of the epidemic, AI has played an increasingly important role
in the fight against the epidemic. Robots can work day and night for disinfection,
which provides a labor force. Robots can make hundreds of calls in a minute for
case screening, which improves the efficiency. Robots can patrol streets and give
adequate warnings, which decrease conflicts. The point is that, in some extremely
dangerous work, robots can come into the forefront, such as direct contact with
patients in quarantine wards, which limits people-to-people and face-to-face contact
and saves epidemic control supplies such as masks, clothes, caps, shoes, gloves and
eyeglasses. At the beginning of the epidemic, masks were rare. Therefore, saving
one supply might mean saving a life (Fig. 1).
Other than working in dangerous areas and densely populated areas, robots can
offer more professional services. For example, Shanghai Mobile and a partner code-
velop 5G robots that can not only carry out simple assignments such as reserva-
tions and mask distribution but also care about patients, answer questions, spread
knowledge of epidemic control, offer video interactions, assist patients and doctors
in communicating remotely and help doctors examine patients by videos that are
later uploaded. Furthermore, Shanghai Mobile applies the 5G infrared thermal
imaging temperature testing system to efficiently test people’s temperature in densely
populated areas for screening dangerous individuals.
As one of the leading AI companies in China, iFLYTEK has assisted doctors
with its smart medical system to follow up and screen tens of millions of people
and analyze millions of medical records upon which reports are made as the basis of
relevant departments’ decision-making. The smart teaching system of iFLYTEK has
connected 19 provinces and 31 regions in China, including Beijing, Anhui, Hubei
and Shanghai, covering more than 8 million students.
Upon construction and acceptance, Huoshenshan Hospital and Leishenshan
Hospital, the two famous hospitals of which the construction was monitored by
tens of thousands of people online, started the 5G plus remote diagnosis method.
With the method, multiple remote consultations for patients with emergencies and
severe symptoms were held, and the regional shortage of quality medical resources
was effectively relieved.
6 The Use of AI in Law in Response to Major Epidemics 211

Fig. 1 Map of people movement from Wuhan developed by Baidu

On February 4, 2020, the Ministry of Industry and Information Technology issued

the Initiative on Fully Performing the Empowerment Function of Al and Jointly
Fighting Against the Novel Coronavirus Pneumonia Epidemic to call for using AI
to empower epidemic governance and speed up the application of AI products at the
forefront of epidemic prevention and control. The initiative includes measures such
as scientific research, epidemic control improvement, epidemic control app develop-
ment, mass production of epidemic control facilities, pooling intelligent resources,
expansion of varied online apps, optimization of the algorithm and computing power
of AI and assistance with epidemic medicine development.
The central government calls for a race against the virus and against time, and AI
is the X factor that truly participates in the race. The reasons why AI has superiority
in epidemic control are as follows:
First, AI technology can predict the epidemic and facilitate the targeted imple-
mentation of policies. Examples are alerts on the basis of the statistical analysis
by AI technology, all kinds of warnings made by relevant departments in China
for epidemic control and so on. At the same time, screening cases by big data is
highly targeted and less aimless so that opportunities are seized for targeted policy
implementation.
212 Y. Cui

Second, the utilization of AI technology can decrease the contact between

epidemic control staff and confirmed cases, suspected cases and potential virus
carriers, which reduces the risk of contagion and wins time for the fight against
the epidemic.
Third, AI technology can be applied on a large scale, which saves manpower and
plays a huge role in solving the shortage of epidemic control personnel. AI can be
used in assignments such as epidemic control supply distribution, food delivery and
reservation.
Fourth, analysis reports with AI technology backed by big data are reasonable,
impartial and fair, which reduce rumors and play an active role in calming people
down.
Fifth, AI technology and related applications can replace a large number of offline
works. The state has enacted measures for home isolation, work resumption delay and
school opening delay. Through remote control, online working and smart classrooms,
scheduled work can continue.

1.3 Both AI and Law Are Important for the Fight Against
the Epidemic

Compared with the SARS epidemic in 2003, AI technology was the new force in the
fight against the epidemic. Connected with big data, cloud computing, the Internet,
the Internet of Things and 5G, AI technology is broadly applied in smart disinfec-
tion, smart screening, pedestrian alert and analysis, blocking and quarantine, ques-
tion answering and remote consultation. High-tech products such as robots, drones
and infrared equipment play an important role in densely populated areas such as
airports, stations and hospitals. As nonstop professional workers, they keep frontline
medical workers away from sources of infection, which effectively reduces the risk
of cross infection and improves the quality and efficiency of the fight against the
epidemic. Furthermore, we can also see the application of AI technology in foreign
epidemic control systems. Intelligent technologies such as temperature measuring
robots and hospital mobile disinfection robots have been gradually used at the front-
line of epidemic control. In addition, some Western research institutions have used
big data and computer simulations to simulate medicine development for epidemic
control and clinical trials, which has sped up medicine development.
AI is a new technology nevertheless. Relevant technologies have shown initial
maturity; however, back to society, AI has shortcomings in regard to legal norms,
technology ethics and individual rights. Additionally, there are some problems
concerning the use of AI technology. For example, medical institutions, virus
research institutions and researchers unreasonably used the data and overempha-
sized publishing papers; some people felt resentment because relevant depart-
ments published their information about residence, transport and hotels and caused
some conflicts; in addition, some information collecting agencies illegally sold the
6 The Use of AI in Law in Response to Major Epidemics 213

collected personal information and so on. All of the above expose the lack of legal
norms, ethical norms and policy norms for AI.
The state has attached great importance to industry development and the normative
application of AI. In October 2018, the Political Bureau of the Central Committee of
the Communist Party of China organized an education on AI. General Secretary Xi
Jinping stressed that AI was an important driving force of the new scientific revolution
and industrial changes and that accelerating the development of the new generation
of AI was a strategic issue deciding whether China could seize the opportunity for the
new scientific revolution and industrial changes. At the same time, he also pointed
out that we should establish a system of laws, regulations, ethical norms and policies
related to AI to promote the healthy development of AI. The COVID-19 epidemic
makes us realize not only the necessity of speeding up the development of AI but
also the room for improvement in the proper utilization of data, personal information
protection and privacy protection concerning AI technology. Thus, it is necessary to
accelerate the implementation of the AI strategy and enact more laws and regulations
related to AI.
As technology advances, many aspects of society change, especially epidemic
control, where AI technology can be used to carry out assignments that used to require
much time. The application of AI technology can significantly improve epidemic
control, reduce costs and popularize the application of laws. For epidemics, AI and
law have preliminary respective advantages, and the combination of the two can
significantly increase the efficiency of solutions to epidemic-related problems and
effectively remedy shortcomings of past epidemic control. Moreover, the improve-
ment of the legal system related to AI also deserves more attention. At present, China
has already placed enough attention on AI technology but not enough attention on
the combination of technology and law. Therefore, policies on the application of AI
in the field of rule of law shall be strengthened to promote the combination of AI and
law so that both of them can better serve epidemic control and social development.
Although AI is not as intelligent as people imagine, the combination of AI, big
data, cloud computing, the Internet, the Internet of Things and 5G has played an irre-
placeable role in the fight against the epidemic. Behind them are the silent dedication
of countless researchers, designers and engineers who work days and nights to be our
support in the fight against the COVID-19 epidemic. Like the rest of the State, they
respond to the appeal of the Party and the State, they are not afraid of hardship, they
are one of hundreds of millions of heroes of China in the fight against the epidemic
and they are the important force of the great renewal of the Chinese nation. When
we call for the rapid development of the AI industry and the establishment of legal
norms of AI, we pay them our compliments.
214 Y. Cui

2 Application of AI in the Fight Against the COVID-192

At the twelfth session of the Reform Committee of the Central Committee of the
Communist Party of China, General Secretary Xi Jinping stressed that digital tech-
nologies such as big data, AI and cloud computing were encouraged to be used to
better underpin epidemic monitoring and analysis, virus origin tracing, epidemic
prevention, control and treatment and resource allocation. On February 4, 2020,
the Ministry of Industry and Information Technology issued the Initiative on Fully
Performing the Empowerment Function of Al and Jointly Fighting Against the Novel
Coronavirus Pneumonia Epidemic to call for further empowerment of AI, organizing
scientific research and production and prioritizing the development and application
of products relevant to epidemic control.
The new generation of AI, as a new technology extends and expands human
intelligence, has also faced a test in the fight against the epidemic, and what is the
result? By demonstration in multiple application scenarios, AI has underpinned the
fight against the epidemic. There are 5 application scenarios as follows.

2.1 Rapid Screening Realized by Highly Precise AI

Noncontact Temperature Measurement

In COVID-19 epidemic control, it is imperative for every resident to take temperature

measurements when entering public areas such as communities, subways, malls and
office buildings. On the afternoon of February 10, 2020, when General Secretary
Xi Jinping inspected and instructed COVID-19 epidemic control in Beijing, he also
received a temperature measurement when entering the Anhuali community, Anzhen
subdistrict, Chaoyang district. Epidemic control is a responsibility, and measuring
temperature has already become part of the daily routine of everyone during epidemic
prevention.
Teachers in some schools measure students’ temperature from 6.a.m every day,
collect and tabulate the data, report the data to grade and school leaders and then
inform parents. In particular, property management practitioners measure the temper-
ature of anyone who enters communities, and some practitioners even visit every resi-
dents’ house to test the temperature. This kind of manual temperature measurement
is arduous. Additionally, the distance between the person who performs the temper-
ature measurement and the person who receives the measurement is too short, and
especially in most communities, there is only one thermometer, which significantly
increases the risk of cross infection.
In the fight against the COVID-19 epidemic, highly precise AI noncontact temper-
ature measurements have played an important role because of their high efficiency and

2Author: Wang Chunhui, professor, doctor and chief researcher in Institute of China ICT
Development & Strategy, Nanjing University of Posts and Telecommunications.
6 The Use of AI in Law in Response to Major Epidemics 215

precision. For example, there is a highly precise AI noncontact temperature measure-

ment equipped with an infrared temperature measurement module that enables it
to precisely measure temperature without contact. The author notices that, in some
areas, the AI contact temperature measurement plan combines body recognition plus
facial recognition plus infrared/visible light dual sensors, which support noncontact
temperature measurements from distances over 3 m. This kind of plan is mainly used
in densely populated scenarios such as railway stations, bus stations, subway stations
and airports to assist staff in rapidly identifying fever patients. Although people wear
masks and hats, rapid screening can be realized, and the recognition error is within
0.3 °C.
The author also notices that some hospitals adopt a 5G-based temperature
measurement and patrol system that can not only precisely measure the tempera-
ture of a person within 10 m by infrared but also recognize whether people wear
masks. For a person whose temperature is above the alert number or who does not
wear masks properly, the system will sound a temperature warning and a reminder
with high pitch.

2.2 Diagnosis Efficiency Significantly Increased by Medical

Imaging AI

Currently, the case confirmation of COVID-19 relies on the virus nucleic acid test.
The test is highly specific but less sensitive. In addition, according to current expe-
rience, we know that the disease manifests itself on pulmonary imaging earlier
than clinical symptoms. Therefore, radiologic tests are irreplaceable in preclinical
screening. However, doctors have to carefully review over 300 CT images before
confirming a case, and reviewing images by humans is too subjective. If there is not
enough information to use, it is easy to give incorrect judgments. A doctor needs
to examine dozens of patients, which is highly stressful. By learning a number of
medical images, AI can help doctors locate the nidus, reduce missed diagnoses and
misdiagnoses, screen highly suspected patients and reduce the queuing time for
patients and the risk of cross-infection among the staff.
For example, Alibaba has developed a new set of AI diagnosis technologies for the
clinical diagnosis of COVID-19 and has conducted the first test in Qiboshan Hospital
(Zhengzhou’s version of Xiaotangshan Hospital), Zhengzhou. The technology can
deliver precise judgment on the CT images of a suspected COVID-19 case within 20 s,
and the judgment accuracy is 96%. Therefore, the technology can qualify the severity
of symptoms and remarkably increase the diagnostic efficiency. Another example
is that Haidian Hospital, as a designated hospital for the treatment of COVID-19
in Beijing, is the first to introduce the image AI pneumonia diagnosis assistance
system, which can deal with 300 chest radiographs in 10 s and assist the diagnosis
and quantitative evaluation of suspected COVID-19 cases so that the diagnostic
efficiency has been significantly increased.
216 Y. Cui

2.3 Burden of Medical Staff Lightened by Intelligent Robots

There was a viral article named the Incomplete Construction Manuel of Huoshenshan
Hospital that demonstrates the hardship of the construction of Huoshenshan Hospital
and Leishenshan Hospital from a panoramic perspective, starting from the presen-
tation of data including the construction period, the installation process of medical
equipment and logistic supplies. The manual mentions a kind of medical intelligent
robot jointly donated by Cheetah Mobile and OrionStar. The kind of robots has
varied functions and forms. Some robots for guidance can perform guidance without
human operation, automatically respond to ask patients about fever, guide patients
and give them an elemental diagnosis, and help doctors to make remote diagnoses.
Some robots are mobile delivery robots that can deliver test sheets and medicine.
Robots are used to replace medical staff to do simple but laborious repetitive tasks.
To prevent cross infection, quarantine wards, ICUs, operation rooms and fever
clinics of hospitals in Wuhan and Shanghai use intelligent disinfection robots that can
work 24 hours a day and remarkably protect the safety of medical staff and lighten
their workload. The author notices that robots deliver meals and goods in quarantine
areas; for example, meal delivery robots can provide quarantine areas with noncontact
necessity delivery services, such as meals. Staff only have to place meals on the three-
tier tray of the robot at the meal time, input the room number in the quarantine area
on the screen, and automatically deliver meals to the destination. In this time, the
smart delivery robot of JD Logistics also plays a good role. On February 6, 2020, the
robot completed its first delivery in Wuhan, successfully delivering medical supplies
and necessities from the Renhe office of JD Logistics in Wuhan to Wuhan Ninth
Hospital.

2.4 Epidemic Services Supported by AI Smart Voice

Consultation Systems

In the fight against the COVID-19 epidemic, AI smart voice technology has provided
multiple voice consultation epidemic services. For example, the smart epidemic robot
developed by Alibaba DAMO Academy can answer questions about the epidemic,
matters concerning seeing a doctor and protective measures by AI technologies such
as voice recognition and natural language understanding to prevent the shortage of
medical resources and reduce the risk of cross infection. Baidu has built a smart
calling platform for epidemic control, which can provide calling services in three
scenarios: screening nonlocal residents, screening and following local residents and
informing specific population groups. By the platform, 1,500 calls can be made in one
second. Some AI companies have released smart voice epidemic robots that provide
governments and medical institutions with free services such as screening and follow-
up by AI calls. Users can set the calling time. These robots are mainly used to follow
up residents, sent informative texts to residents, screen nonlocal residents, monitor
6 The Use of AI in Law in Response to Major Epidemics 217

health, provide protective suggestions, record the epidemic of the key population
groups and conduct periodical follow-up.
In the epidemic, some hospitals have provided AI smart screening and guidance
services that can rapidly give suspected patients diagnosis and care advice. The
author has attempted to use the AI smart screening and guidance service in Nanjing
Drum Tower Hospital. First, we searched Nanjing Drum Tower Hospital on WeChat
and followed it; then, we clicked the patient service button on the menu bar and
found the online hospital; and finally, we clicked the COVID-19 smart screening
tool on the top. After that, there will be an AI doctor, and the patient only needs
to talk about his symptoms and answer questions on suspected contact history. The
AI screening and guidance service system will immediately make judgments and
offer treatment solutions to prevent patients from blindly going to the hospital and
remarkably decrease cross-infection.
The author also notices that Raisound, Mediway and FoundAI have developed
an intelligent voice HIS system and an elevator voice control module, which help
decrease the risk of people getting the COVID-19 twice because of contact. Smart
customer service robots/assistants developed by Yun Wen Technology, Yiducloud,
China Life Financial Insurance and Antgroup have functions such as smart epidemic
notification and follow-up, key population group follow-up and expert inquiry assis-
tance. It can answer users’ questions related to epidemic control and provide them
free online doctor service.

2.5 Epidemic Tracing System Built by AI to Quickly Trace

Transmission Pathways

It is crucial for COVID-19 epidemic control to identify the population groups classi-
fied as high-risk and superspreaders in a timely manner, obtain precise information
on the development of the epidemic, rehearse different epidemic policies and eval-
uate the effect, and quickly trace the transmission pathway of the virus so that the
control efficiency of government is enhanced. In the epidemic, the author knows an
AI-based plan for precise control, epidemic prediction and virus tracing. AI tech-
nology was developed by 4 Paradigm, a Chinese AI company, in collaboration with
Nanjing University and clinical experts of Northern Jiangsu People’s Hospital. It is
an AI-based plan for precise control, epidemic prediction and virus tracing, and now
it is officially used on the frontline of the fight against the epidemic.
In epidemic control, it is crucial to precisely screen highly risky and vulner-
able population groups. By identifying the population group and adopting measures
such as timely discovery, notification and quarantine, the epidemic can be effec-
tively controlled. The traditional screening rule and system is based on the judgment
of whether a person has simultaneously appeared with the confirmed or suspected
patient in the same area. It is simple and easy to follow. However, it is particularly
218 Y. Cui

difficult to obtain the data sources to be analyzed. For the accuracy of the screening
results based on limited data sources, there is considerable room for improvement.
AI-based plans for precise control, epidemic prediction and virus tracing use AI
technology to enrich the current epidemic screening rule and model, further expand
the screening coverage, improve the callback rate and accuracy, and decrease the
difficulty and cost. Particularly when it is emergent, the plan can assist the compe-
tent department in investing limited resources in an important area. Moreover, the
technology uses machine learning to establish a data-driven COVID-19 transmission
digital twin system, set up the potential transmission network, depend on the infor-
mation of patients and identify the potential transmission source and the potential
superspreader to help the epidemic control department quickly control the source.
Actually, there are three reasons why the new generation of AI truly works: data,
computing power and algorithms. Computing power computes data. The algorithm
establishes distinctive models for different industries. Only by combining three of
them are there scenarios for AI technology because the foundation of AI is training.
Similar to humans, if they want to acquire a skill, it must be acquired through contin-
uous training. Thus, for AI, considerable high-quality data are vital, and the data
should be related to all possible scenarios so that it is possible to create an excellent
mode. The algorithmic training and deep learning of AI require a great amount of
data. However, at present, there are few samples of confirmed cases at the beginning
of the epidemic, and medical institutions do not have high-quality clinical data. With
the accumulation of clinical data, the big data-based radiologic features of COVID-
19 will be gradually clearer, and the algorithm and the computing power of AI will
play a more important role in the diagnosis of the disease.
The author notices that in the COVID-19 epidemic, most AI technology companies
not only actively provide AI technology and services but also share the computing
power and algorithms of AI. Compared with the application that needs to meet the
demand of specific scenarios, the computing power of AI plays a universal role in
the fight against the epidemic. AI technology companies announce the sharing of
their own computing power, and they support epidemic control with action.
On January 29, 2020, the Alibaba Cloud officially announced that it would share all
AI computing power for free and provide AI computing power to support virus gene
testing, drug development and protein screening to help research institutions shorten
the R&D cycle. On January 30, Tencent officially announced that it would provide
free supercomputing power, cluster computing support and basic cloud computing
power for Professor Huang Niu’s laboratories at the National Institute of Biolog-
ical Sciences, Beijing and at the Interdisciplinary Institute of Biology and Medicine,
Tsinghua University and Professor Luo Haibin’s team at the School of Pharmaceu-
tical Sciences, Sun Yat-Sen University. On January 30, Baidu announced that it would
share for free the linear fold algorithm, which was a linear-time algorithm, and the
fastest RNA structure prediction website in the world, which shortened the prediction
time of the secondary RNA structure of the novel coronavirus to 27 seconds, with
genetic testing institutions, epidemic control centers and scientific research centers
around the world. On February 1, Didi Cloud also joined the ranks by sharing with
domestic scientific research institutions, medical and rescue centers for free GPU
6 The Use of AI in Law in Response to Major Epidemics 219

cloud computing resources and technical support for the work related to the fight
against the epidemic. Generally, AI has played an active role in the fight against the
COVID-19 epidemic, and its empowerment function has been fully performed.

3 Application of Health QR Code and Legal Adjustment

in Epidemic Prevention and Control3

When a health QR code greatly helps support epidemic prevention and control,
it also brings problems such as personal information infringement. A Health QR
code is a colorful QR code that is generated upon risk evaluation according to the
health condition in specific scenarios and by the e-account of personal health cards
established by an administrative department through personal identity symbols or the
unified health symbols of organizations based on personal health cards. In essence,
a health code is a digital pass or health certificate used in certain ranges with the
coordination of government.

3.1 Technical Architectures of Health QR Code

Depending on how they are generated, there are centralized health codes and
distributed health codes. Additionally, the two types differ in the information
collection range.

3.1.1 Distributed Health QR Code

When using distributed health code, an individual needs to install the app on his
phones. The app uses Bluetooth to locate the phone or uses an ultrasonic method to
detect whether there are individuals with the same app passing by. If there are such
individuals, the app will record them.4

Architecture of Distributed Health QR Code

The distributed health QR code uses Bluetooth’s BLE (Bluetooth Low Energy) broad-
casting technology, which enables devices to broadcast information to other nearby
Bluetooth devices via the BLE protocol.

3 Author: Chen Jidong, lecturer, Law School, Shanghai University, S.J.D.; Hao Yuxuan, assistant,
Research Center for Data and Laws Shanghai Jiao Tong University, postgraduate.
4 See Zhu Yue, Analysis of Global Health Code Ethical Dilemma: Three Obstacles to Algorithmic

Governance, reproduce from Tengyun (腾云), an official account on WeChat.

220 Y. Cui

Every device generates Ephemeral IDs. An Ephemeral ID relies on some cryp-

tographic algorithm to be generated, and therefore, it needs a private key. For the
purpose of security, the private key needs to be changed every day, and the private
key of the day is generated based on some hash of the private key of the previous
day. To further protect the anonymity of Ephemeral ID, the design requires multiple
new Ephemeral IDs to be generated every day to alternate according to the corre-
sponding time interval. The device has a local time setting for data expiration so that
the expired data will be automatically deleted.5
If a patient is confirmed, the process will be as follows. First, the patient actively or
passively (depending on the country’s policy) uploads the information of private key
in the first day being infective to the back-end server. After receiving the information,
the back-end server broadcasts the private key information to all devices using the
server by pushing notifications or periodically polling and scraping. When the phone
receives private key information, it does a calculation locally to evaluate the infection
risk. If the risk is above the threshold, the app will give an alarm. A close contact
whose risk is above the threshold can choose to upload contact information, such as
desensitized anonymous data, such as times and periods, to the disease prevention and
control department or epidemiologists. The contact can also choose not to upload
the information. After the confirmed patient completes the upload of private key
information, a brand-new private key will be generated at random, ensuring that his
privacy will not be violated afterwards.6

Features of Distributed Health QR Code

The distributed health QR code focuses on free will and privacy protection. It does
not collect phone location data, mobile profiles, contact information or identifiable
features of terminal devices. Therefore, it is more like a digital measuring tape that
warns the user when getting too close to the infected. The back-end server is also only
a communication medium that itself does not store and analyze data on a large scale.
Even if the server is hacked or abused, the risk of privacy violation can be minimized.
Users have the option of whether to share personal information with relevant staff of
the disease prevention and control center. Data shared by users have been anonymized
and desensitized before the disease prevention and control center receives them, so
it is impossible to acquire detailed location and real identity information from the
received data.7

5 Apple and Google, Privacy-Preserving Contact Tracing, www.apple.com/covid19/contacttracing/.

6 DT3P White Paper, Decentralized Privacy-Preserving Proximity Tracing, 12th April 2020. https://
github.com/DP-3T/documents/find/master?q=.
7 See Fang Jian, Different European and American Health Codes, reproduce from Shujuguan, an

official account on WeChat.

6 The Use of AI in Law in Response to Major Epidemics 221

3.1.2 Centralized Health QR Code

Based on the centralized data storage and analysis model, the centralized health
code uses backstage-connected databases and information filled by individuals to
make comprehensive color judgments. Currently, all existing health codes in China
are centralized health codes, including the anti-epidemic health information code
launched by the national integrated government service platform and health codes
structured into provinces and cities.

Architecture of Centralized Health QR Code

At the public end, the entrance mostly relies on the existing real-name platform
such as Alipay, WeChat and DingTalk. There are other entrance options, such as
using PC apps or developing exclusive mobile apps. The user end is based on real
data filled out voluntarily, including information such as address, health condition,
questions whether or not you have been to the infected area or whether or not you
have contacted key people.
At the end of the analysis, there were two kinds of coding data resources. One
is the personal information filed by the people end voluntarily. The other is the
data gathered by the state, departments and regions, including data of confirmed or
suspected cases from the national health department, data of close contacts in the
same train or planes from the departments of exit-entry, civil aviation and railway,
and travel data for the last 14 days from telecom operators. Both kinds of data are
uploaded to the centralized data storage and analysis platform of the government.
The national anti-epidemic control and health information code depend on the CTID
platform, while other health codes developed by provinces and cities depend on
different IT platforms to compare and analyze data and then generate health codes
of different colors.
At the offline end of prevention, control and test, prevention and control testers
require people to scan QR codes or show health codes. If necessary, measures such
as temperature measurements can be used. In this way, quick inspection and testing
can be achieved in various scenarios (Fig. 2).

Features of Centralized Health QR Code

The generation of green, orange and red codes is mainly based on three dimensions.
The first is the space dimension, which means that the data judgment includes towns
and townships/subdistricts in accordance with the risk degrees of epidemics across
the country. The second dimension is the time dimension, which means the times and
periods a person has been in the infected area. The third dimension is the relationship
dimension, which means contact with key people. Upon analysis and evaluation based
on epidemic control rules and data modeling, three risk statuses can be measured.
The red code represents high risk, the yellow code represents medium risk and the
222 Y. Cui

Fig. 2 Technical architecture of epidemic prevention postcode8

green code represents healthy status or law risk. It is dynamic among the three color
codes.
Information included for analysis differs from region to region. Generally, the
analysis covers information such as the answer to question whether or not you have
been confirmed or suspected, the recent activity trace and areas and the recent flight
or train. Some regions also collect address information and the answer to question
whether you have bought fever medicine. As the epidemic and the backstage of
the health code develop, the collection scope of information for analysis changes
dynamically.

3.2 Application of Health QR Code in the World

3.2.1 Distributed Health QR Codes Adopted by Europe and America

The California Consumer Privacy Act (CCPA) and the General Data Protection
Regulation (GDPR) strongly protect and regulate data and privacy, fully consid-
ering the influence of data and privacy protection. Hence, in full consideration
of the influence of data and privacy protection, most European and American
countries chose distributed health codes when designing the epidemic surveillance
plan. On April 10, 2020 local time, Apple and Google announced that they would
jointly use Bluetooth technology to help governments and health agencies contain

8Shenzhen Standards Promotion Council, Shenzhen, Reference Architecture and Technology Guide
of Anti-epidemic Pass Code, T/SZS 4010–2020.
6 The Use of AI in Law in Response to Major Epidemics 223

the epidemic. They launched public health apps and gradually included the tech-
nology into operating systems in a direct way. Approximately 3 billion iPhones and
Android phones can wireless exchange anonymous information. Germany has also
adopted a similar architectural model and developed CoronApp under Pan-European
Privacy-Preserving Proximity Tracing (PEPP-PT).

3.2.2 Centralized Health QR Codes Broadly Used in China

Health QR Codes Used in 34 Regions in China

On February 11, 2020, Hangzhou first released the health QR code, and by February
17, Zhejiang became the first province in China to have full health code coverage.
On February 29, the national integrated government service platform launched a
nationwide anti-epidemic health information code, which can be used in 34 regions
in China, including Hong Kong, Macau and Taiwan. Regions such as Zhejiang also
demand returned ethnic Chinese and overseas Chinese, foreign students and other
foreign nationals to apply for the international version of the health QR code.
However, from the data published by provinces, the population coverage of health
codes of different provinces differs significantly. Table 1 shows the use of health codes
in the 16 provinces with officially published data.

Diverse Application Scenarios of Health Codes

During epidemic control, health QR codes have been widely used in traffic manage-
ment, residential area management, management of work units including enterprises,
public institutions, universities, middle and elementary schools, public area manage-
ment and other fields. As long as people scan or show the health QR code and their
temperature is normal, traffic can be normal, work and production can be resumed and
school can be opened. For risk alert management, health codes can be used to iden-
tify and analyze risks to public health security and alert the public. Administrative
departments can perform assignments such as regional epidemic surveillance, health
monitoring and interdisciplinary research on key population groups and targeted
delivery of health education. By using health codes to collect relevant data and infor-
mation, infectious disease surveillance models of time, space and population groups
are established to enable alerts to public health emergencies. For the response to
the epidemic emergency, health QR codes are used in epidemiological investiga-
tions, cluster case investigations and quarantine of sources of infection and close
contacts. Meanwhile, big data generated in the application in multiple scenarios
strongly support epidemic prevention and control.
Apart from epidemic control, attempts to connect health codes and medical
systems and establish regional integrated medical and health systems based on health
codes are also underway. On May 22, 2020, Hangzhou announced that it would launch
gradient color health codes with which users will know their health condition in the
 224

Table 1 Use of health QR codes in some provinces in China

Zhejiang Shandong Hainan Tianjin Anhui Jilin Inner Mongolia Hunan
Number of health codes issued 88.151 ma 55.67 mb 5.027 mc 8 md 21.460 me 8.6776 mf 5.30 mg 14 mh
Permanent resident population in 2019 58.50 m 100.7021 m 9.4472 m 15.6183 m 63.236 m 26.9073 m 25.396 m 69.1838 m
Population coverage (%) 150.7 55.3 52.3 51.2 33.9 32.3 20.9 20.3
Guangxi Liaoning Henan Guizhou Ningxia Chongqing Sichuan Hebei
Number of health codes issued 10 mi j 13.98 mk 5.20 ml 648 km 2.74 mn 3.33 mo p
8.677615 m 2.072 m
Permanent resident population in 2019 49.60 m 43.517 m 96.40 m 36.2295 m 6.9466 m 3.12432 m 83.75 m 75.9197 m
Population coverage (%) 20.1 19.9 14.5 14.1 9.3 8.8 4.0 2.7
a As of April 13, 2020, Zhejiang province has issued a total of 88.151 million health codes, of which 57 thousand are international versions. b As of May 4, 2020,

Shandong province has issued a total of 55.67 million health codes. c As of February 26, 2020, Hainan province has issued a total of 5.027 million health codes.
d As of April 14, 2020, Tianjin has issued a total of 8 million health codes. e As of March 17, 2020, Anhui province has issued a total of 21.46 million health
codes. f As of April 9, 2020, Jilin province has issued a total of 867,7615 health codes. g As of May 4, 2020, Inner Mongolia autonomous region has issued a total
of 5.30 million health codes. h As of April 2, 2020, Hunan province has issued over14 million health codes. i As of April 3, 2020, Guangxi Zhuang nationality
autonomous region has issued over 10 million health codes. j As of April 10, 2020, Liaoning province has issued a total of 8.6776 million health codes. k As of
April 12, 2020, Henan province has issued a total of 13.98 million health codes. l As of March 2, 2020, Guizhou province has issued a total of 5.20 million health
codes. m As of March 12, 2020, Ningxia Hui nationality autonomous region has issued a total of 648 thousand health codes. n As of March 2, 2020, Chongqing
has issued a total of 2.74 million health codes. o As of February 25, 2020, Sichuan province has issued a total of 3.33 million health codes. p As of April 13,
2020, Hebei province has issued a total of 2.072 million health codes
Y. Cui
6 The Use of AI in Law in Response to Major Epidemics 225

future and that by the data integration of e-medical records, physical examination and
lifestyle management, it would explore the establishment of personal health index
rankings based on related health index and colors of health code. In the future, health
codes will still be needed to help prevent future epidemics, promote the innovation
of the health service model, and build smart cities.

3.3 Existing Problems in the Application of Health QR Codes

3.3.1 Problem of Coding Accuracy

Some of the data sources of health QR codes are the information filled in by the people
themselves. Concealment and misreporting increase the difficulty and workload of
information comparison and affect epidemic prevention and control, so it is necessary
to focus on how to ensure the authenticity of information filled by people themselves.
In addition, there are more urgent problems to be solved in the application of health
QR codes: the problem of accuracy and lag between the data scrapped by backstage
and the reality of people, the problem of how to provide equivalent verification
methods for individuals unable to use health QR codes, the algorithmic accuracy
problem and the problem of how to handle appeals.

3.3.2 Problem of Nationwide Interconnectivity and Recognition

Health QR codes developed independently by different regions are still not recog-
nized nationwide. The phenomenon that a person holding multiple codes has to
submit information on different apps every day to go to work normally has caused
inconvenience to the public. Thus, it is a major issue for normalized epidemic control
that is how we promote nationwide interconnectivity and recognition of health QR
codes so that people can truly travel around the country with one code on hands.

3.3.3 Problem of Personal Privacy Protection

Articles 1032 and 1034 of the Civil Code stipulate that citizens have the right of
privacy and that personal information is protected by laws. In the process of using
health QR codes, much personal information is involved, such as name, ID number,
contact information, location, travel and health. If the information is not properly
stored, disclosed or abused, information security will be compromised, and social
ethical issues such as discrimination due to sensitive characteristics such as regional
features may also arise.
226 Y. Cui

3.4 Solutions to the Problems

3.4.1 Improvement of Coding Accuracy

Improvement of Data Matching Accuracy

Because the different big data engines used by different provinces are different,
the accuracy of data scrapped by backstage differs. Technically, real-time or retro-
spective analysis can be applied to analyze the relationships among humans, time
and space to achieve accurate temporal and spatial information. It is significant
for visualized epidemic prediction, targeted warning for close-exposure groups
and epidemiological investigation of population groups, especially close-contact
population groups.

Dynamic Adjustment and Risk Evaluation Standard Criteria

As the epidemic is in dynamic change, the risk evaluation criteria and the coding
principle should be dynamically adjusted based on the development of the epidemic
and in response to the local demand for work resumption to fit actual health and safety
conditions and meet the needs of people in life. At the same time, the algorithm should
be promptly adjusted based on relevant feedback to ensure the maximum fairness
and accuracy of the algorithm.

Establishment of a Punishment Plan for Dishonest People

To ensure the accuracy of self-reported data, punishment for dishonest people should
be initiated. Those who do not report personal information truthfully shall be clas-
sified as dishonest people as punishment. Administrative punishment should be
imposed on serious cases. Those who have caused certain consequences shall be
prosecuted for crimes related to obstruction of infectious disease control.

Establishment of an Appeal Review and Investigation Mechanism

The generation of health codes is an automatic decision-making mechanism that

significantly affects the rights and interests of information collecting subjects. In
accordance with the Personal Information Security Specification (2020 Edition),
personal information collecting subjects can complain of the automatically decided
result and request a manual review.
For the situation in which the coding does not match reality, an open and trans-
parent procedure should be established to receive and promptly handle appeal-related
health codes and provide timely feedback, and 24-h hotlines and online feedback
6 The Use of AI in Law in Response to Major Epidemics 227

channels should be set up. Backup plans shall be offered for special situations, and
emergency plans for unexpected situations shall be formulated.

Paper Health QR Code

With regard to individuals who do not use smartphones, health codes can collect only
the data of confirmed or suspected cases from health departments and other existing
data, such as data of close contacts sharing the same railways or airplanes with
confirmed or suspected cases, but not the real-time location data of the individuals.
A family member can apply on behalf of an individual a paper health code that shall
be carried out when passing through traffic checkpoints and entering and leaving
communities, schools and public areas and scanned by working personnel who record
his current location. If the individual is confirmed to have a health risk through the
time and space information, he shall be timely informed through backstaging.

3.4.2 Problem of Nationwide Interconnectivity and Recognition

Formulation of a National Unified Standard

On March 5, 2020, the draft of the Reference Architecture and Technology Guide of
the Anti-epidemic Pass Code was officially submitted for approval, which provides
technical specifications for anti-epidemic pass code developers and prescribes
the functions required for personnel management, authorization management and
collection point management of pass code using units.

Establishment of a National Data Sharing Platform

On February 29, 2020, the national integrated government service platform released
a national anti-epidemic health information code. Health codes of different regions
can be connected with the anti-epidemic health information code, which is used
as a medium to transform themselves so that local health codes can be recognized
in other regions. Regions without local health codes can directly use the national
anti-epidemic health information code and combine relevant information about local
epidemic prevention and health to achieve interconnectivity and mutual recognition
in different regions.

3.4.3 Problem of Personal Privacy Protection

The Law on Prevention and Treatment of Infectious Diseases prescribes that no

disease control and prevention institutions and medical institutions shall disclose rele-
vant data and information on personal privacy. The Notice on Effectively Protecting
228 Y. Cui

Personal Information and Using Big Data to Support Joint Prevention and Control
issued by the Office of the Central Cyberspace Affairs Commission specifically
prescribes that the collection of personal information for joint prevention and control
shall follow the Personal Information Security Specification, which is a national stan-
dard and adhere to the principle of the minimum extent. The Personal Information
Security Specification (2020 Edition) requires that for information systems with
automatic decision-making mechanisms, the evaluation of the effects on personal
information security shall be carried out at the stage of planning and design or before
the first use and effective measures to protect personal information collecting subjects
shall be taken based on the evaluation results.

Regulation of Information Processing Ways

In nature, the generation of health codes is the processing of personal information.

When collecting information, information collecting subjects shall be informed by
specifically informing users in a conspicuous position that data required for health
codes are collected by the government or other departments. Prior to the personal
information collection, it shall be made clear that the information to be collected and
used will be only for the work related to health codes, what personal information
will be directly and indirectly collected and what information of the user will be
extracted when the health code is scanned, and the express consent shall be obtained
from the user. With specific targets and the minimum principle, only the minimum
information that can achieve the purpose of epidemic prevention should be collected.
Security and confidentiality measures should be taken when storing and trans-
mitting information. Personal information preservation shall be handled by security
measures such as corresponding secure encrypted storage. Personal information shall
not be expressively encoded in the health code, sensitive personal information shall
not be directly encoded in the health code by ways like simple transcoding, and
both of them shall be encrypted. There shall be staff designated for information
processing who shall be informed of the security responsibility. Permission manage-
ment for checking data shall be used, and the data sharing shall be limited in the
scope of local epidemic control requirements and the coding principle.
With regard to the analysis and use of information, the collected basic information
shall be used for the architecture of health codes, not for other purposes. Only the most
necessary part of the data that can be used to quickly trace the close contacts shall be
preserved. We can learn from the distributed health code, which anonymizes data and
uses techniques such as encryption, verification and access control and cryptographic
measures to protect privacy, prevent unauthorized scanning clients from accessing
health code information and deidentify sensitive personal information after the health
code is scanned and before the content appears.
The algorithmic logic of the health code shall be open, transparent, clear and
interpretable. Simple algorithmic logics shall be directly disclosed, while models
difficult to explain shall be accompanied by accurate and understandable explanations
of the workings. To avoid discrimination based on sensitive features and because
6 The Use of AI in Law in Response to Major Epidemics 229

most health code discrimination is about regional features, the health code system
shall instantly update the notification of risk level in each place and exclude the
regions where the risk of epidemic is already low in the judgment criteria. Under
no circumstance shall the criteria include sensitive features such as gender, race
and nationality. It should be guaranteed that scanning clients and institutions that
provide health codes with technical services and support do not hold health codes and
the related personal information and alter the purpose of use without authorization.
Without the user’s consent, no organization or person shall disclose his health code
information.
The collected information shall not be unlimitedly stored. In the distributed health
code, data will be deleted after the fixed time. We can learn from this kind of code to
set the validity term of the health code so that when the validity term ends or the user
actively stops using the health code, the related personal information will be promptly
deleted or handled in accordance with laws, and the time and space information of
the user whose health is guaranteed and the time and space information exceeds the
isolation period will be deleted in a timely manner.

Introduction of Stricter Regulation and Punishment

The regulation department shall fully implement measures related to data security
and personal information protection and prevent violations such as the disclosure and
abuse of data. It will impose more severe punishment, handle violations in a timely
manner, such as the disclosure and abuse of health code data, publish typical cases
to society and improve the deterrent effect of the cases.

4 Role of Online Direct Reporting System in Epidemic

Prevention and Control and Legal Regulation9

The China Information System for Disease Control and Prevention (hereinafter
referred to as the Online Direct Reporting System) is the national surveillance
information system platform for disease control and prevention connecting all-level
health administrative departments, health units and relevant organizations and is a
comprehensive application and information system platform established in the field
of national disease prevention and control to collect, organize, store, search, analyze
and research information, make decisions and provide information services. The
Online Direct Reporting System was established after the SARS epidemic in 2003,
with nationwide coverage in April 2004. On January 24, 2020, a dynamic surveil-
lance function for pneumonia caused by the novel coronavirus was launched on the
National Notifiable Diseases Reporting System. Laws and regulations specifically

9Author: Chen Jidong, lecturer, Law School, Shanghai University, S.J.D.; Hao Yuxuan, assistant,
Research Center for Data and Laws Shanghai Jiao Tong University, postgraduate.
230 Y. Cui

stipulated the use of the Online Direct Reporting System for online direct reporting
epidemic information of infectious diseases and public health emergencies.

4.1 Working of the Online Direct Reporting System

4.1.1 Architecture of the Online Direct Reporting System

The core subsystem of the Online Direct Reporting System is the National Notifi-
able Diseases Reporting System (NNDRS), which enables the real-time, online and
direct case reporting of statutory infectious diseases based on health institutions and
has functions such as data collection, real-time statistical analysis, timed statistical
analysis and visualized presentation based on the geographic information system
(GIS).
The Online Direct Reporting System uses the B/S (Browser/Server) architecture
and adopts the working mode where browsers make requests and the server responds.
Hospital uses the browser end, while the server end is deployed in the data center
of Chinese Center for Disease Control and Prevention (hereinafter referred to as
the CCDC) so that it is guaranteed that all community hospitals only need to use
a browser to enter the system and report cases, without the need for step-by-step
reporting and step-by-step approval, and cases can reach the data center of CCDC
directly.
The system uses a virtual private network (VPN) built on the Internet to transmit
data. The gateway of VPN enables remote access by encrypting data packets and
converting packet destination addresses so that data security is guaranteed.
In April 2008, the CCDC started the operation of the China Infectious Diseases
Automated-alert and Response System (CIDARS) in 31 provinces (including
autonomous regions and municipalities directly under the central government)
nationwide and established an outbreak automated alert and response mechanism.
CIDARS uses fixed threshold and time models to automatically analyze infectious
disease surveillance data reported by 31 provinces nationwide based on the Online
Direct Reporting System in terms of county (district) and on a daily (or real-time)
basis. It sends alert signals of abnormal disease increase or aggregation to epidemic
surveillance personnel of corresponding county (district) centers for disease control
and collects the responses to all alert signals. At present, it has functions for 39
infectious diseases, such as automated analysis of surveillance data, real-time identi-
fication of spatiotemporal clustering, sending of alert signals and real-time follow-up
of responsive results. Figure 3 demonstrates the specific architecture of the Online
Direct Reporting System and the CIDARS with the participation of all systems.
6 The Use of AI in Law in Response to Major Epidemics 231

Fig. 3 Specific architecture of the online direct reporting system and CIDARS10

4.1.2 Reporting Procedure of the Online Direct Report System

In terms of data reporting, after confirming infectious disease cases, clinicians fill
infectious disease report cards semiautomatically or manually and submit them to
the hospital’s public health department or administration department of nosocomial
infection, who will input the cases into the Online Direct Reporting System. By then,
real-time cases will be directly reported to the National Individual Cases Data Center
of CCDC.
For the reporting procedure, cases are reported to the district or county center for
disease control to be verified and included in the single/specific disease management
information system, and then the cases are reported to the municipal and provincial
platforms in turn and finally sent to the NNDRS via the national health data exchange
platform. The four mentioned public health data exchange platforms improve the
format and completeness of data. Centers for disease control and prevention at all
levels simultaneously report cases to health administrative departments at the same
level, and a mutual feedback mechanism is formed among health administrative
departments at all levels.
With regard to the surveillance end, there are designated staffs in CCDC to monitor
data and write daily analysis reports. Once more than 5 cases of unknown pneumonia
are found in a single location, a verification mechanism will be automatically trig-
gered, and the CCDC will send staff to conduct epidemiological investigations, visit
patients and take samples.

10See Ushier, Hardcore: Improving National Infectious Disease Surveillance and Alert Network
with Blockchain Technology, reproduce from Everyone is a Product Manager, an official account
on WeChat.
232 Y. Cui

The Online Direct Reporting System has the following features: The first is direct
reporting. Hospitals directly upload infectious disease report cards to the direct
reporting system without reports and approvals level by level, and the reporting
cannot be intercepted. The second one is individual cases reporting. What are reported
are individual cases including medical information other than privacy, rather than
the data reporting with monthly summary data. The third one is instant reporting.
Reporting in the system is immediate so that data are received by CCDC within 4 h.

4.2 Application of Online Direct Reporting System

4.2.1 A Well-Developed and Broadly Applied System

The Online Direct Reporting System in China is the largest online direct reporting
system for epidemics and public health emergencies in the world, which is across the
board. Disease prevention and control institutions above the county level achieved
100% infectious disease online direct reporting and shortened the reporting time
from 5 days to 4 h. The online reporting rate of medical institutions above the county
level reaches 96%, the number of community medical institutions is 85%, and the
underreporting rate for infectious rate continues to decline.11
The system obtains data from an increasing number of sources. Since 2004,
CCDC has established several single-disease surveillance systems, such as tuber-
culosis management information systems, plague prevention and control manage-
ment information systems, and AIDS comprehensive prevention and control infor-
mation systems. Some single-disease surveillance systems can have mutual access to
NNDRS through standard interfaces to push data of individual cases. To prevent the
repeated entry of epidemic data from medical institutions and improve the complete-
ness and accuracy of surveillance data, the trial connection between the Online Direct
Reporting System and the Hospital Information System (HIS) was initiated in 2007 to
realize the semiautomatic submission of infectious disease report cards through auto-
matic pop-up of the medical records or manual opening of the filling page. In 2011,
a trial based on regional population and health information platforms to exchange
epidemic surveillance data from residents’ electronic health records (EHRs) and
electronic medical records (EMRs) was launched, a unified collection and exchange
platform for public health data was planned, and information platforms at the provin-
cial, municipal and county levels were connected with the Online Direct Reporting
System.
In regard to coverage, the Online Direct Reporting System covers all disease
prevention and control institutions above the county level, as well as 98% of the

11See Wang Chen: Inspection Report of the Law Enforcement Inspection Team of the Standing
Committee of the National People’s Congress on the Implementation of the Law of the People’s
Republic of China on Prevention and Treatment of Infectious Diseases, at the 5th session of the
Standing Committee of the 13th National People’s Congress on August 28, 2018.
6 The Use of AI in Law in Response to Major Epidemics 233

medical institutions above the county level and 96% of the health centers at the
township or town level in China.12 In terms of categories of infectious diseases, China
has achieved real-time and online monitoring of 39 statutory infectious diseases.
It has also been proven that in response to epidemic emergencies such as H5N1
virus, human diseases of Streptococcus suis, H1N1 virus, Chikungunya virus and
H7N9 virus, China has actively and effectively controlled the further development
of epidemics by using the Online Directly Reporting System.

4.2.2 Limited Application at the Beginning of Epidemic, but More

Effect Later

At the beginning of the COVID-19 epidemic, the direct reporting system did not
work as an application. On January 20, 2020, when the State Council classified novel
coronavirus pneumonia as a Class B infectious disease, to apply the preventive and
control measures for a Class An infectious disease, it was the first time the Online
Direct Reporting System was used to report cases. Take Guizhou as an example.
On January 21, it began to report information about novel coronavirus pneumonia
cases on the Online Direct Reporting System. On January 24, the CCDC released the
dynamic surveillance function for pneumonia caused by the novel coronavirus on the
NNDRS. Since then, the Online Direct Reporting System has actively played a role
in supporting epidemic data filling and level-by-level statistics, covering information
such as suspected and confirmed cases and continuously improving the quality and
efficiency of data reporting, so that cases are reported within 2 h after discovery and
tertiary review is completed within 2 h.

4.3 Existing Problems of the Online Direct Reporting System

at the Beginning of the Epidemic

4.3.1 Unreasonable Designs of Infectious Disease Report Cards

At present, the national infectious disease report card is triggered based on the ICD-10
diagnosis code to report 39 known infectious diseases. COVID-19 was not recognized
as a statutory infectious disease at the beginning of the epidemic. Therefore, it could
not be uploaded to the statutory infectious disease system. For this kind of disease, the
design of infectious disease report cards in Guangdong, Inner Mongolia and other
regions included the choice of reporting pneumonia of unknown etiology, while
the infectious disease report card in Wuhan did not have such a choice, which led
to the dilemma of no confirmation, no filling and directly prevented doctors from
conducting timely online reporting.

12See Xu Wen: The Epidemic Reporting System Behind the New Pneumonia, published on Beijing
News, https://www.sohu.com/a/370630234_114,988.
234 Y. Cui

4.3.2 Flaws in the Reporting Procedure for Pneumonia of Unknown

Etiology

The COVID-19 epidemic could have also been reported through the pneumonia of
unknown etiology (PUE) direct report subsystem. In accordance with the National
Plan for Surveillance, Screening and Management of Cases of Pneumonia of
Unknown Etiology, a case that meets the definition of PUE should be reported imme-
diately to the relevant authorities upon discovery, and the medical institution should
arrange an expert team to consult and investigate the etiology. If the etiology is
not clear, the infectious disease report card should be filled out, indicating PUE
and directly reported to the online system. This process is parallel to the reporting of
infectious disease report cards, which closes the loopholes in the design of infectious
disease report cards and allows for online direct reporting throughout the country.
However, centers for disease control did not give enough training, and the original
purpose of the Online Direct Reporting System was to monitor infectious diseases
such as SARS and the lack of epidemics for years caused doctors and medical insti-
tutions to feel no more alert. As a result, most doctors did not know the definition
of PUE cases or the PUE reporting system and did not understand that the reporting
procedures would affect case reporting.
In addition, before the reporting of a new disease of PUE by the system, it was
imperative to first review the report by the hospital and then to arrange an expert
consultation. Only when the experts determine the disease or are still unable to make
a clear diagnosis can the medical institution report the case by the Online Direct
Reporting System. Experts take a long time to diagnose, evaluate, and verify the
disease, resulting in low efficiency and long reporting cycles for PUE surveillance
in local hospitals.

4.3.3 Alert System Only Effective for Known Diseases

Currently, the epidemic alert system in China is not a conditional model based on
big data analysis but a rule judgment model based on the results of infectious disease
report cards. The rule judgment model makes judgments and alerts only for known
diseases, so it does not work for new infectious diseases of unknown etiology.

4.3.4 Lack of a Horizontal Information Sharing Mechanism Among

Regions

The reporting mechanism in the disease prevention and control system is managed
vertically and territorially in a hierarchical way, which means that there are national,
local, county, township and town levels and centers for disease prevention and control
at each level share information with and give feedback to the health administration
departments at the same level. Hospitals do not have the comparison data of patients
with the same symptoms. Infectious disease report cards are only uploaded vertically,
6 The Use of AI in Law in Response to Major Epidemics 235

with no horizontal information sharing. Alerts and warnings to other hospitals only
rely on the China Infectious Diseases Automated-alert and Response System, which
causes the alert means to be too limited to make prompt responses.

4.4 Suggestions for Improving the Online Direct Reporting

System

4.4.1 Online Direct Reporting System Optimized by Blockchain

Technology

Establishment of a Disease Control Chain of Four Levels Including District,

Municipality, Province and State

The current vertical and one-way infectious disease reporting network shall be opti-
mized by using the blockchain sharing mechanism to establish a blockchain (disease
control chain) automated data synchronization network of four levels including
district, municipality, province and state and relying on centers for disease control at
all levels to improve the abilities of emergent infectious disease data collection and
real-time alter and self-governance.
Using the existing public health data exchange platforms as the switching nodes
for data of each level, a real-time automated data exchange mechanism shall be
established. Infectious disease report data of districts will be synchronized in the
municipal epidemic prevention chain to synchronize the national epidemic prevention
data in the national chain. When the four-level disease control chain automatically
completes the internal alert for epidemic prevention in all districts, municipalities
and provinces, it keeps updating and adding data of other provinces by the national
disease control chain to form an anti-epidemic network with a certain degree of
self-governance capacity. Therefore, the problem that other provinces cannot obtain
information on the reported cases can be solved.13

Establishment of a State-Community Dual Alert Network

A set of big data conditional analysis models shall be built in the infectious disease
alert system as the supporting library for offline prediction and alerts to collect risky
symptom data of infectious diseases and Internet data in addition to infectious disease
report cards. For example, a national top-level infectious disease offline alert network
shall be established by using information such as flight ticker reservation, medicine
supply and search data. Meanwhile, based on the real-time data synchronization of the

13➀➁➂Ushier, Hardcore: Improving National Infectious Disease Surveillance and Alert Network
with Blockchain Technology, reproduce from Everyone is a Product Manager, an official account
on WeChat.
236 Y. Cui

four-level disease control chain, real-time data judgment and alert competence based
on rule models shall be developed by smart blockchain contracts. The current national
infectious disease automatic alert function shall sink into epidemic prevention chains
at the district, municipal and provincial levels to form a real-time community alert
network.

Establishment of an Infectious Disease Reported Data Sharing Network

Hospitals and centers for disease control (CDC) at the district (county) and municipal
levels are the basic channels to report infectious disease cases. A distributed and
peer-to-peer infectious disease data sharing network shall be established to enable
the automated synchronization of epidemic data among hospitals and regions and
to realize cross-regional and cross-level data exchanges in the four-level disease
control chain through four-level health data exchange nodes. A single chain realizes
data synchronization within the region through the blockchain automated node data
synchronization capability.

Establishment of an Accountability and Storage Mechanism for Reported

Data

Anti-alteration and transparent blockchain technology can provide a complete

accountability system and offer a data basis for transparent supervision and account-
ability. Patient reception, handling of suspected cases and confirmed cases, death
and reports can be under the supervision of the public and health administrative
departments. Once it is necessary to investigate accountability for major epidemics,
a complete and anti-alternation accountability chain can be formed based on the data
traceability of the blockchain.

4.4.2 Establishment of a Diversified Infectious Disease Surveillance

Mechanism

Apart from the Online Direct Reporting System, a diversified surveillance mech-
anism shall be established, the public shall be encouraged to identify and report
infectious diseases and the public awareness of infectious disease surveillance and
alert shall be heightened by legislation and education. Let disease prevention and
control institutions, medical institutions and blood collection and supply institutions
and their executive functionaries, health administrative departments at all levels, mili-
tary medical institutions, ports, airports and railway disease prevention and control
6 The Use of AI in Law in Response to Major Epidemics 237

institutions, and border health and quarantine authorities participate in the disease
control chain of epidemic detection and reporting.14

4.4.3 Improvement of the Content Design of Infectious Disease Report

Cards

Infectious disease report cards are essentially conclusive data, so it takes a long time
to artificially determine suspected and confirmed cases. After a big data-based condi-
tional alert model is established, in addition to the collection of infectious disease
report card data, we can also extract risky symptom data of infectious diseases from
medical records and examination reports, such as fever, cough, chest radiograph
descriptions and biochemical indexes, separate and identify key labels of risky symp-
toms of infectious diseases by unstructured data analyses and upload the labels auto-
matically through the four-level health data exchange nodes to establish an online
direct reporting system with symptoms at its core.

4.4.4 Simplification of the Procedural Limitations on Epidemic

Detection Reporting

Following the principle of incentives, institutions, organizations and individuals in

communities should be encouraged to promptly input epidemic information into
the disease control chain upon detection. Relying on all nodes on the disease control
chain, epidemic information collection, analyses, research and verification should be
simultaneously conducted. All secret information related to epidemics (such as the
privacy of patients) shall be encrypted by private keys, while all nodes on the chain
can read only public information related to epidemics by public keys and promptly
make decisions based on respective duties, functions and authorities. The decisions
shall be synchronized in the chain to ensure that the work of epidemic prevention
and control is always ahead of the spread of the virus.15

4.4.5 Establishment of an Information-Sharing Platform

for Professionals

Based on the disease control chain, a quick reporting and professional support plat-
form should be built among medical practitioners. On the platform, professionals
can share and exchange information in a timely manner, make timely warnings and

14 See Dong Liwu: A Few Suggestions on the Amendment of the Law on the Prevention and Control
of Infectious Diseases with Blockchain Thinking, reproduced from Shanghai Law Society, an official
account on WeChat.
15 See Dong Liwu: A Few Suggestions on the Amendment of the Law on the Prevention and Control

of Infectious Diseases with Blockchain Thinking, reproduced from Shanghai Law Society, an official
account on WeChat.
238 Y. Cui

set up risk levels at the outbreak of the epidemic to which other regions across the
country can quickly respond. Meanwhile, diagnosis experiences and methods can
be exchanged across the country, and high-quality medical resources can be shared
online.

4.4.6 Implementation of Mechanisms of Training, Supervision,

Evaluation and Punishment

Centers for disease control shall organize regular training for frontline medical prac-
titioners to ensure that accountable staff know how to use the Online Direct Reporting
System to accurately, efficiently and promptly conduct the online direct reporting of
infectious diseases. The centers shall also include emergency management-related
training in the daily work of community public health practitioners.
Information must be objective and true, the reported content must be accurate and
comprehensive, and relevant factors must be normative and complete. Evaluation
indexes shall be determined, and the evaluation of online direct reporting in all
regions shall be made on a regular basis.
Health administrative departments at all levels shall implement the punish-
ment mechanism. For infectious diseases confirmed through channels outside the
Online Direct Reporting System that have caused serious consequences, the relevant
personnel should be held accountable.

5 Establishment of an Information Disclosure Mechanism

of Epidemic Accountability16

5.1 Issue Raised

5.1.1 Basic Rights and Social Demands for Information Disclosure

of Epidemic Accountability

Citizens’ knowledge of epidemic information and supervision of epidemic preven-

tion and control will influence authorities when making anti-epidemic decisions.
Thus, to protect citizens’ right to know about public epidemic prevention, it is neces-
sary to provide citizens during the epidemic with quick and true information about
epidemic prevention and control that meets their demand. Among the information,
the disclosure of information on administrative accountability for epidemic preven-
tion and control is an important way to encourage citizens to engage in the supervision
of government departments, medical institutions and other epidemic controllers as
well as a crucial link in pushing the structured information disclosure of public health

16 Author: Bao Kun, Ph.D. Candidate, Ko Guan Law School, Shanghai Jiao Tong University.
6 The Use of AI in Law in Response to Major Epidemics 239

emergencies. Whether all kinds of information about negligence in epidemic preven-

tion and control can be focused on or followed-up by the public greatly depends on
the channels, scope and level of information disclosure and the quality of infor-
mation. Among them, some accountability information disclosure with significant
social influence and public attention is vital for citizens to assess public epidemic
control during special times, which means that information disclosure is one of the
key factors and triggers of public sentiment.17

5.1.2 Obstacle to Information Disclosure

The government is an organization strictly following existing rules in operation. Thus,

when the COVID-19 epidemic breaks out with highly uncertain content, outlines and
boundaries, the work mechanism of the public authority formed in the normal period
is less likely to respond quickly.
Based on the conflict between the reasonable demand for information disclo-
sure and the obstacle to information disclosure, the article collected and organized
information from the Internet and found that during the fight against the epidemic,
administrative accountability information disclosure had elements such as nonstan-
dard formality, unclear classification and inconsistent calibers. It is necessary to,
through the statistics, untangling and organization of information, find reasonable
categorization methods and promote the establishment of a standard information
disclosure mechanism.

5.2 Sources of Samples and Data

Currently, research on government information disclosure is general on disclosure in

nonepidemic periods; however, accountability for epidemic prevention and control
during the epidemic period is distinct in terms of its subject, content and result,
and it is difficult to collect relevant information in nonepidemic periods for in-depth
research. Therefore, by means of a data collection method that has collected epidemic
accountability information from the Internet and public service platforms in different
regions, this study has acquired 1,779 pieces of accountability information published
from January 21 to March 5, 2020 by data scraping and artificial organizations. To
study citizens’ right to know, data were obtained from sources that are accessible to
ordinary citizens on the Internet, including websites, official accounts on WeChat and
Weibo accounts of discipline inspection commissions and supervisory commissions
at all levels across the country. To ensure the completeness and accuracy of data,
this article initially used scraping software to scrape high-frequency words such as
epidemic, accountability and inefficiency on the websites and official accounts, then

17See Liu Zaichun, Analysis of the Online Anti-Corruption Information Disclosure System, Hubei
Social Sciences, No. 1, 2014.
240 Y. Cui

reduced noise data and finally completed the structuring and labeling of the data
through artificial intelligent semantic recognition tools under human supervision so
that it formed a multidimensional information base including subjects, content and
time of accountability as well as facts, reasons and results of violation, which is the
reference framework to unify the calibers of diverse information in the epidemic
accountability information disclosure mechanism and provides a basic reference for
the establishment of a stable information disclosure mechanism.

5.3 Status of Information Disclosure of Epidemic

Accountability and Pathway to Standardization

5.3.1 Framework of Epidemic Accountability Information Disclosure

Basic Logic of Information Disclosure

Accountability is a broad system that requires specific scenarios to determine the defi-
nite accountability subjects and identifiable targets and give clear evaluations of the
accountability subjects, which means that the logic of epidemic information disclo-
sure shall be consistent with the specificity of prevention and control work during the
epidemic period. Meanwhile, although it is the statutory obligation of government to
disclose information, disclosure is not the same as sharing. Constrained by adminis-
trative efficiency and cost, the disclosure of epidemic accountability information must
be limited to the extent of meeting the public’s demand to know epidemic prevention
and control, which means that the disclosed information shall be both sufficient and
necessary. Information also shall be disclosed in a logical and comprehensive way so
that the quality of disclosure can be guaranteed. The disclosure subject of account-
ability information shall disclose information on the subject, object and content of
accountability. Information on accountability subjects concerns the issue of who
shall hold accountable, which can help citizens initially judge whether the subject
has timely dealt with the dereliction of duty in epidemic prevention and control as
required by its duty. The accountability object information focuses on who shall be
held accountable, which helps the public focus on defaulters and their dereliction of
duty who are around citizens or exposed by the Internet or news and highlights the
participation of the public in accountability. The content of accountability is about
what accountability, which can enable the public to judge based on all kinds of facts
whether or not the results are appropriate, and guide the public to engage in the
substantive judgment of accountability.
6 The Use of AI in Law in Response to Major Epidemics 241

Analysis of Data Completeness in Each Link of Information Disclosure

In terms of the completeness of accountability data of the epidemic information

disclosure (see Table 2), the information on accountability subject and object with
identification function is fairly complete, and the completeness of information on
subject, open channel, regions to which the accountable person is affiliated, organi-
zation to which the accountable person is affiliated and his post, illegal action and
accountability result is 100%. However, there are some weaknesses in the disclosure
of accountability information. For example, first, a few accountability announce-
ments do not include accurate accountability time (86 pieces, or 4.8% of the total
pieces), and more accountability announcements do not specify the exact time when
the violation occurred (612 pieces, or 34.4% of the total pieces). The significance
of the coexistence of two data points is to judge the overall speed of the discovery-
investigation-treatment process of the accountability authority for the violation and
dereliction of duty by comparing the interval between the occurrence time of the
violation and dereliction of duty and the accountability time. Second, there are
inconsistent and nonstandard names of a few accountability objects in accountability.
Specifically, the names of individuals held accountable in 16 pieces of data have been
changed to alias or alternative names, and the reasons for the anonymization are not
given. The information disclosure system is restricted from excessively disclosing
the privacy of accountable individuals; however, in the emergent period of epidemic
prevention and control, for citizens to perceive the accuracy and effectiveness of
accountability, it is generally inappropriate to disclose accountability information in
an anonymous manner. If it is necessary not to disclose information to protect the
privacy of the accountable individuals, proper reasons shall be specified in announce-
ments. In addition, the information under the category of fact of violation of account-
ability content shall be improved. There were only 310 pieces of data (37.5% of the
total data) mentioning that there were consequences of violations and dereliction
of duty, which can be understood as only a small number of violations and derelic-
tion of duty has caused certain consequences, but the expression of consequences is
also quite general and abstract: among them, 120 announcements describing conse-
quences have used only abstract expressions such as “caused serious impact” and
“caused great passivity”. Based on such information, it is difficult for the public to
judge the causality between consequences of actions and severity of accountability,
which is not conducive to the public judging whether accountability follows the
principle of proportionality.

5.3.2 Accountability Subjects

The eligibility of accountable subjects in epidemic prevention and control is also a

major link to the information disclosure mechanism, which should be a focus. In
epidemic times, accountability is more urgent than in normal times, but account-
ability for epidemic prevention and control can only rely on general accountability
legal systems, party regulations or provisional policies, which greatly increases the
242 Y. Cui

Table 2 Framework of information disclosure and the completeness of data in each section
Information category Completeness of information
(No. of missing pieces)
Information on accountability Subject 100%
subject Time 95.2% (86)
Public channel 100%
Information on accountability Region 100%
object Name (individual, company, 99.2% (16)
organization)
Organization affiliated 100%
Position 100%
Information on accountability Fact of violation Time 65.6% (612)
content Action 100%
Consequence 37.5% (1469)
Accountability result 100%

complexity of accountability subjects and causes ambiguity in the nature of account-

ability subjects in information disclosure. To prevent too many kinds of subjects from
participating in the accountability, which will reduce administrative efficiency and
increase the legitimacy risk after the accountability, the public authority tends to adopt
a monistic method in practice, authorizing only one subject as the core force to hold
accountable. It has been proven that China used the same method during epidemic
prevention and control. According to all the data collected in this study, as shown
in Fig. 4, party subjects18 accounted for 97% of the total accountability subjects,
while administrative subjects as the sole subject19 accounted for only 3%. Among
the Party subjects, discipline inspection commissions and supervisory commissions
are the majority, accounting 93%, while only 4% are local party committees of
governments. Based on this approach, the problem of efficiency and risk of substan-
tive accountability in epidemic prevention and control were reduced; however, the
simple structure of accountability centralizes the information and power, which will
lead to the singularity of the subject of accountability information disclosure, repel
other subjects who have the right to participate in and supervise information disclo-
sure. The simple structure lacks the openness and supervisability that reflect the value
of the procedure and is inevitably questioned by the public about neutrality.
In a special period, if the simple structure is chosen for substantive accountability
for efficiency reasons, the way to compensate for the lack of neutrality in the proce-
dure is by the greater diversity of the government information disclosure subjects

18 The Party subjects mentioned here include both Party organizations at all levels, as well as super-
visory organs subordinate to Party leadership groups, such as discipline inspection commissions
and supervisory commissions.
19 The administrative subjects mentioned here mainly refer to governments at all levels and their

departments as well as public institutions engaged in epidemic prevention and control.

6 The Use of AI in Law in Response to Major Epidemics 243

Fig. 4 Proportion of accountability subject types

in the information disclosure. On the one hand, information disclosure shall be led
by governments and party organizations at all levels. According to the attributes of
China’s political system, the party and government serve as the ruling body and
administrative core of China’s political system, respectively, and are supposed to
shoulder the leading responsibility and the direct responsibility to promote account-
ability information disclosure as well as supervise governments at a lower level to
perform the basic work of information disclosure. On the other hand, according to
the new Regulations on Government Information Disclosure, channels for informa-
tion disclosure should be opened up for multiple supervisory subjects in society
under the leadership of the Party and the government, and people’s congresses at
all levels, judicial organs and the public have the right to supervise the effectiveness
and timeliness of government information disclosure. A structure with one leading
party and multiple supportive parties is conducive to establishing an accountability
system of balance and mutual promotion, which effectively increases the neutrality
of accountability information disclosure.

5.3.3 Categories of Actions Subject to Accountability

From the perspective of the relationship between legal norms and social facts, the
accountability mechanism in special periods is similar to what Habermas made in
his famous book Between Facts and Norms that there is an internal tension between
objective facts and social norms, and the effectiveness of norms depends on their
reflection of objective facts.20 In the special scenario of accountability for epidemic
prevention and control, the substantive effectiveness and the public acceptance of
accountability will greatly rely on whether regulations can provide reasonable feed-
back to old, new, various and intricate violation and dereliction of duty in reality. To
avoid the problems of accountability, such as dysfunction, inaccuracy, high arbi-
trariness and generalization, which are caused by the mechanical application of
general accountability regulations in normal time to the dereliction of duty in special
time, it is necessary to conclude the types of actions subject to accountability in the

20See [German] Jürgen Habermas, Between Facts and Norms, translated by Tong Shijun (2003
version), SDX Joint Publishing Company.
244 Y. Cui

existing accountability information from the perspective of pure facts. According to

the sample data, the existing actions subject to accountability are categorized by the
following frameworks (Table 3):

Table 3 Table of categories and number of actions subject to accountability

Category of actions subject to accountability No. of pieces
(proportion)
General prevention and Intentional dereliction of Absence from post 186 (10.4%)
control duty without permission
Nonfeasance 65 (3.4%)
Negligent dereliction of Inadequate general 1013 (56.9%)
duty community control
Missed testing 16 (0.8%)
Information Information reporting Omission and 20 (1.1%)
transmission misreporting
(negligence)
Concealment and 42 (2.3%)
delay in reporting
(intentionality)
Information disclosure Release of inaccurate 13 (0.7%)
information
Information leakage 22 (1.2%)
about epidemic
prevention
Inappropriate remarks 47 (2.6%)
Mismanagement of Ineffective control of people who have been in 35 (1.9%)
special individuals key epidemic areas
Ineffective management of suspected cases 10 (0.5%)
Ineffective patient control 9 (0.5%)
Individual violation of Gathering Gathering for 33 (1.8%)
laws and discipline drinking
Gathering for 50 (2.8%)
gambling
Gathering for general 137 (7.7%)
activities
Trouble making 13 (0.7%)
Obstruction of law enforcement 22 (1.2%)
Mismanagement of Abuse of power for personal gains 9 (0.5%)
epidemic prevention Mismanagement of medical supplies 3 (0.1%)
supplies
Price gouging of medical supplies 4 (0.2%)
Leadership responsibility 50 (2.8%)
6 The Use of AI in Law in Response to Major Epidemics 245

General Prevention and Control

Communities are the frontal battlefield of the fight against epidemic, so this type of
accountability mainly focuses on the dereliction of duty of epidemic prevention and
control staff in communities when directly managing the public epidemic preven-
tion activities in communities in accordance with the deployment of their superiors,
emphasizing the direct connection between accountable individuals as well as their
superiors and all communities under management.

Information Transmission

The prompt and accurate transmission of epidemic information21 is urgently needed

in the epidemic prevention and control period, which is directly related to the effec-
tiveness of epidemic prevention and control. However, as community epidemic
prevention and control highly requires information transmission by humans and is
limited by technologies and human factors, it is still difficult to prevent dereliction
of duty, such as untimely, inaccurate and untrue epidemic information transmission.

Mismanagement of Special Individuals

The management mechanism for special individuals is the basis of the prevention and
control of highly contagious epidemics, but it is also a high-risk task that is most likely
to cause a significant risk of virus transmission due to inefficient management. There-
fore, it is necessary to place more attention on the supervision and accountability of
the task.

Individual Violation of Laws and Discipline

In the special time of epidemic prevention and control, prevention and control workers
(especially party members) shall not only complete their duties but also comply with
the anti-epidemic policies stipulated for the special time and laws and regulations in
force from time to time as citizens. The violation of laws and discipline of epidemic
prevention and control workers will compromise the power of epidemic prevention
and control teams and even trigger a crisis of public truth.

21 Epidemic information mentioned here shall be understood in a broad sense, referring all necessary
information for the epidemic prevention and control, including but not limited health information
and travel history in key epidemic areas of residents, information on suspected cases and information
on patients, etc.
246 Y. Cui

Leadership Responsibility

Accountability for epidemic prevention and control is more severe than general
accountability. On the one hand, the system in which leaders are responsible for
epidemic prevention and control shall be reasonably implemented to share the risk
of community executors to prevent the unfair phenomena of letting the executive
positions take the blame to cope with the supervision of superiors or public pressure
from dampening the enthusiasm of prevention and control workers.22 On the other
hand, accountability shall be targeted. Using appropriate measures to deal with offi-
cials who are responsible for the incident can not only deter and educate themselves
but also others. Leadership responsibility is generally dependent on other specific
types of actions subject to accountability; however, it is not deniable that the super-
vision and instruction of leaders has certain independence. Hence, it is included as a
separate category of information as a reference for the public to monitor the behavior
of leaders.

5.3.4 Accountability Results

Accountability Terms and Their Frequency

Based on the results of data collected from the Internet, a number comparison
chart of accountability results is made according to the terms used in accountability
announcements.
The information disclosure of accountability results is the most concerned part
when the public supervises dereliction of duty in epidemic prevention and control,
which represents that the final evaluation of violation and dereliction of duty by
accountability authorities influences the public’s behavior in the epidemic.23 There-
fore, the results of accountability shall be not only substantively in line with the
value orientation of epidemic prevention and control but also in accordance with
laws and regulations. Accountability within the statutory scope shall be made by
the authority with the power of accountability. Although Fig. 5 show that there is
an inverse relation between the severity and the number of accountability results,
terms used in announcements are quite general and not uniformly normative, so it is
hard to directly distinguish the source and nature of punishment from the contents of
announcements, which makes it impossible to judge whether accountability subjects
held people accountable as required by laws.

22 See Jiang Laiyong, Research on the Formation and Correction Mechanism of “Accountability
Alienation”, Henan Social Sciences, No. 7, 2019.
23 For example, in some widely influential accountability cases occurred in the nationwide fight

against the global pandemic, such as the case of Dali seizing masks, etc., the action by the authority
holding the negligent staffs of such cases accountable will significantly influence the public and
epidemic control departments at all levels in the way of handling the epidemic-control supplies and
in the tolerance of unverified epidemic information released by bona fide medical workers.
6 The Use of AI in Law in Response to Major Epidemics 247

Fig. 5 Chart of the number of accountability results in announcement terms

Normative Classification of Accountability Results

Categories of accountability results and corresponding authorized accountability

subjects shall be strictly in compliance with laws to avoid arbitrary and simpli-
fied accountability. In accordance with competent laws (see Table 4), the results of
accountability shall be categorized in terms of the nature of power into three types,
including accountability of administrative officials, intraparty accountability and
liability. In terms of the effects, among the three mentioned accountability methods,
the accountability of supervision organs led by party organization is the most rapid
and deterrent, which can promptly respond to most derelictions of duty in the special
period of a serious epidemic and immediately punish them. Thus, it has a strong
deterrent effect.
After combing Fig. 5; Table 4 and clarifying the categories and sources of power
under the framework of laws, it can be seen that there are six problems of normativity
in the terms of collected accountability data: (1) talk, may include the warning talk
and the reminder talk made by the supervisory authority for a civil servant; (2)
criticism, may include the criticism circulation and the criticism and education made
by the supervisory authority for a civil servant; (3) circulation, may include the
circulation made by a party organization for a party cadre or an inferior organization
and the criticism circulation made by the supervisory authority to a civil servant;
(4) admonition, may include the admonition made by the supervisory authority for
a civil servant and the admonition made by a party organization for a party cadre;
(5) resignation order, may include the resignation order made by the supervisory
248 Y. Cui

Table 4 Categories of accountability results in terms of laws and regulations

Punishment and disciplinary Punishment for civil Warning talk
actions for administrative servants (accountability Criticism circulation
officials subject: supervisory
authoritya ) inspection
Resignation order
Reminder talk
Criticism and education
Inspection order
Admonition
Disciplinary actions for Warning
public officials Demerit
(accountability subjects:
supervisory authorityb , the Serious demerit
affiliated organizationc ) Demotion
Removal from office
Dismissal
Intraparty accountability Intraparty disciplinary Warning
actions Serious warning
(accountability subject:
party organizationd ) Removal from Party post
Probation within the Party
Expulsion from the Party
Accountability for the party Inspection
organization Information circulation
(Accountability subject:
e
party organization ) Reshuffle

Accountability for party Information circulation

cadres Admonition
(Accountability subject:
party organizationf ) Organizational adjustment or Position adjustment
organizational discipline Resignation order
(Subject of recommendation
right: supervisory authorityg ) Removal from office
Suspension
Liability Crimes of dereliction of Negligence
duty Abuse of power
(Accountability subject:
Judiciaryh ) Dereliction of duty of epidemic prevention and control

a See Article 45 of the Supervision Law of the People’s Republic of China; Article 9 of the Provisional Regulations on
Administrative Punishment for Public Officials. b See Article 45 of the Supervision Law of the People’s Republic of
China; Article 6 of the Provisional Regulations on Administrative Punishment for Public Officials. c See Articles 61 and
62 of the Civil Servants Law of the People’s Republic of China. d See Article 8 of the Regulations of the Communist
Party of China on Disciplinary Punishment. e See Article 8 of the Regulations on Accountability of the Communist
Party of China. f See Article 8 of the Regulations on Accountability of the Communist Party of China; Article 54 of
the Regulations on the Selection and Appointment of Party and Government Leading Cadres. g See Article 9 of the
Provisional Regulations on Administrative Punishment for Public Officials. h See Articles 397 and 409 of the Criminal
Law of the People’s Republic of China
6 The Use of AI in Law in Response to Major Epidemics 249

authority for a civil servant and the resignation order made by a party organization
for a party cadre; (6) inspection, may include that the supervisory authority orders a
civil servant to conduct inspection and that the party organization orders an inferior
organization to conduct inspection.
It is clear that there is ambiguity between the source and the type of power in
the accountability results in accountability information disclosure. Such ambiguity
is related to the singularity of accountability subjects, as shown in Fig. 3. In terms
of the ways massive accountability announcements are published, tacitly believing
that accountability results are made by party organizations and that the supervisory
authority is not conducive to accurate and normative epidemic information disclo-
sure. Therefore, accountability subjects must take the language stipulated by law as
the standard to standardize the form of accountability information disclosure and
improve the quality of information disclosure.

5.4 Conclusion

After analyzing the published information on COVID-19 epidemic control, severe

and careful accountability emerged from the sample data. The accountability mecha-
nism has imposed a great pressure of supervision on epidemic prevention and control
workers, directors and decision-makers at all levels, which also demonstrates the high
standard and execution ability of Chinese epidemic prevention and control. However,
even in stressful and special periods, the significance of the epidemic accountability
information disclosure system should not be ignored. A normative and standard
accountability information disclosure mechanism can effectively urge the substantive
accountability procedure to be more in line with laws, prevent a simplified, arbitrary
and abused accountability and indirectly form a benign supervision on the epidemic
prevention and control at all levels. Integrating the value of procedural justice and the
substantive effectiveness required for epidemic prevention and control, this mecha-
nism is an important embodiment of the government pursuing the modernization of
governance capacity and the development of the rule of law during the special period
of epidemic prevention and control.
250 Y. Cui

6 Prospect for Building Justice Big Data to Ensure

Enterprises to Resume Work and Production
in the Postepidemic Era—Focusing on Procuratorial
Work24

Since the outbreak of the COVID-19 epidemic, China has faced serious challenges
to the economy and society. At present, China has made significant progress in
the prevention and control of the epidemic, and the production and operation of
enterprises has gradually returned to normal. However, the resumption still faces
many problems. Especially in the context of economic globalization, enterprises will
be confronted with more new situations and challenges in resuming operation and
production as the global epidemic continues to deteriorate. In this special period, it is
incumbent on the judiciary to provide quality legal guarantees for enterprises to help
them go through the difficult time. This section places attention on the judicial needs
of enterprises, focuses on procuratorial work, comprehensively examines the new
challenges faced by the development of judicial big data in serving the development
of enterprises, and deeply discusses with the epidemic prevention and control and
resumption of operation as the starting point the paths and methods of judicial big data
development, to provide judicial supports and guarantees for the healthy development
of enterprises in the postepidemic era.

6.1 The Need of Judicial Guarantee for Enterprises

to Resume Work and Production

During the epidemic, enterprises may be confronted with financing difficulties,

contract disputes, labor disputes and other problems when resuming operation and
production. Many enterprises face a survival test, and some of them even violate
laws to overcome their production and operation difficulties. As a large number of
conflicts and disputes are to be resolved in judicial ways, the demand of enterprises for
judicial guarantees grows more urgently. The main manifestations are the following
three aspects. There are three aspects of legal guarantees:

6.1.1 More Emphasis on Equal Protection

Equal protection in accordance with the law is the most direct demand and expec-
tation of enterprises for justice. Due to the differences of enterprises in the nature
of property rights, the ability to control capital, technology and other factors of
production and different judicial policies of regions, enterprises are given different

24Author: Qi Yongfu, deputy director, law and policy research office, Shanghai People’s
Procuratorate.
6 The Use of AI in Law in Response to Major Epidemics 251

treatments in litigation. First, many new problems emerge in the economic field,
and the development of laws and regulations lags behind reality. Different judicial
policies in different regions can easily lead to inconsistent standards in judicial prac-
tices so that enterprises may suffer from the problem of judicial injustice of different
judgments for similar cases. Second, handling cases related to enterprises always
requires the organic integration of three effects. From the perspective of ensuring a
stable economy and creating jobs, large enterprises often have more judicial guar-
antees than small and micro enterprises. Moreover, affected by the epidemic, the
economy at home and abroad faces serious challenges, and small and medium-sized
enterprises are under greater pressure to survive. When they are drawn into civil and
commercial disputes or are involved in crimes, their production and operation may
be greatly affected, so they have greater expectations for equal judicial protection.

6.1.2 More Emphasis on Diverse Protection

When facing disputes, enterprises tend to resort to courts for civil, administrative
and judicial remedies, and when violated by illegal crimes, they turn to public secu-
rity organs for criminal and judicial protection. At the time of seeking the above
protection, enterprises may face the problems of economic disputes intervened by
criminal means and false lawsuits, and their legitimate rights and interests may be
violated. In the process of the resumption of operation and production, enterprises
eagerly require more effective remedy channels, including mediation, arbitration and
other diversified dispute resolution mechanisms, as well as through the procurato-
rial organs performing legal supervision functions to correct lax law enforcement,
unfair justice, etc. Taking the procuratorial work as an example. In the new situa-
tion where criminal prosecution, civil prosecution, administrative prosecution and
prosecution for public interest litigation are fully developed in a comprehensive and
coordinated manner, the procuratorate effectively plays the roles of civil, adminis-
trative and public interest litigation prosecution to protect the legitimate rights and
interests of enterprises, which is the new expectation of enterprises for the judicial
protection of procuratorate.

6.1.3 More Emphasis on Efficient Protection

The judiciary case handling efficiency is of vital significance for the survival and
development of enterprises. What enterprises are afraid of about lawsuits is the
time-consuming litigation process, which will trap enterprises in a dilemma. In the
practice of criminal justice, sometimes there is a phenomenon in which when a case is
finished, the enterprise collapses. Other than the failure to normal operation due to the
imprisonment of enterprise leaders caused by inappropriate use of coercive measures,
the traditional and rigid method and low efficiency of case solutions are the most direct
reasons. For civil and administrative disputes that require a long litigation period,
when enterprises are looking for the protection of procuratorial work and supervision,
252 Y. Cui

quick mechanisms of acceptance, handling and feedback mechanisms are important

supports for increasing enterprises’ satisfaction with judicial guarantees. With regard
to the improvement of the quality and efficiency of judicial services, the judicial
branch shall extend the functions of handling cases, strengthen the role of judicial
advice, procuratorial advice, optimize the supply of accurate and efficient judicial
services, etc., which can also help enterprises go through the difficult time.

6.2 New Challenges to Judicial Big Data Construction

in Serving the Development of Enterprises

Influenced by the epidemic, judicial work has changed significantly. Noncontact

remote case handling and online judicial services have been further optimized. The
development of judicial big data confronts both rare opportunities and new chal-
lenges. From the perspective of procuratorial work, the current procuratorial big data
development to serve the optimization of the doing-business environment and the
resumption of business and production still has some room for urgent improvement.

6.2.1 Internal Data System to Be Further Integrated and Improved

Currently, the platform for the development of procuratorial big data is based on
the unified business and application system connecting relevant information systems
and platforms to realize data connection and application in multiple scenarios. With
respect to the whole data system, now, procuratorial organs have a wide variety of
platforms with different data structures, standards, and cleanliness of content, which
has caused the phenomenon of data silos. From the perspective of a unified system,
function modules are designed based on business types, so there is no corresponding
module or rule to extract and aggregate data for block data of specific categories such
as the handling of enterprise-related cases. Considering only the case handling of
criminal procuratorial work, although it is specified that economic crimes are handled
by designated departments, the scope of enterprise-related cases is relatively broad,
among which cases involving persons in charge of enterprises, executives and persons
in charge of other important positions suspected of traffic accidents, drunk driving
and other common criminal offenses are still handled by other departments, resulting
in the decentralization of enterprise-related as isolated data islands. Due to the lack
of data standards and specifications and the knowledge difference of case-handling
personnel, the recording and collection of data of enterprise-related cases may be
impeded, and the systematic and standard pool of big data of enterprise-related cases
fails to be established, which obstructs the realization of big data analyses and relevant
application in different scenarios.
6 The Use of AI in Law in Response to Major Epidemics 253

6.2.2 External Data Connectivity and Sharing to Be Further Improved

Unlike commercial and social data, the level of acquisition and application of legal
big data considerably rely on whether legal organizations fully and timely disclose
the legal information they collect and hold.25 From the perspective of the national
governance system and the modernization of governance capability, the development
of judicial big data cannot be isolated. It must break data barriers and fully connect
external data platforms to realize crossover data collection and sharing. In judicial
practice in Shanghai, with the launch and application of an auxiliary case handling
system for criminal cases (referred to as the 206 System), a unified platform has
been formed for the handling of criminal cases by various organs of politics and
law, which provides a unified channel for the flow of data from all parties, but it
still uses data exchange as the major mode of data flow and integration. In regard
to the performance of procuratorial work and supervision functions, although the
information sharing platform of administrative law enforcement and criminal justice
has now been built up, it is difficult to effectively detect and transfer clues of problems
such as failure to transfer cases and substitution of fines for criminal punishments
because law enforcement data from all parties have not been fully and timely entered.
For procuratorial collaboration in the Yangtze River Delta, there is no cress-region
data exchange channel for procuratorial organs in Shanghai, Jiangsu, Zhejiang and
Anhui, so no unified standard system can be established for the application of law,
evidence examination and case handling procedures related to enterprise-related
cases in the delta, which is not conducive to the resolution of the problem of different
judgments for similar cases in different regions.

6.2.3 Data Mining Application in Scenarios to Be Further Improved

The development of judicial big data is a long-term project in which adequate data
collection is the starting point and foundation and deep mining applications are the
ultimate value. At present, a considerable part of the data on procuratorial organs
are still in a dormant state and are not activated.26 With the rapid development of
information technology and the fast-changing economy at home and abroad, the
development of procuratorial big data shall also advance with time and make the
best of reality to meet market demands and continue improving the deep mining
and application of data. As different regions continue optimizing the doing-business
environment, integrated online service platforms such as Government Online-offline
Platform have become the mainstream products of administrative service, providing
more convenient services to citizens through efficient management of data. Especially
in the epidemic period, an effective communication bridge was erected between

25See Zuo Weimin, Toward Big Data Based Legal Research, Chinese Journal of Law, No. 4, 2018.
26See Ji Meijun et al. Challenges and responses of procuratorial organs in the era of big data,
People’s Procuratorial Semimonthly, No. 15, 2017.
254 Y. Cui

the government and the public and enterprises. In this regard, data on enterprise-
related cases can be analyzed, researched and judged under specific topics to form
analysis reports of specific businesses, provide visualized demonstrations to provide
references for judicial decision-making and case handling, or be connected with
new technologies such as AI to upgrade cloud services to offer procuratiorial public
services of more intelligence and precision for enterprises.

6.3 Prospects for the Path of Deepening the Development

of Judicial Big Data to Facilitate the Healthy
Development of Enterprises

On April 9, the Central Committee of the Communist Party of China and the State
Council issued Opinions on Improving the Systems and Mechanisms for Market-
based Allocation of Factors of Production, which included data as one of the factors
of production for the first time and provided the basic understanding and guidance for
judicial organs to further the development of big data and build high-quality judicial
data assets. In the specific application scenario of serving the healthy development of
enterprises, by regarding epidemic prevention and work resumption as opportunities,
the development of judicial big data can be restructured and improved from three
dimensions: improving the quality and efficiency of judicial case handling, deepening
the comprehensive management of the market and upgrading judicial public services.

6.3.1 Improving the Quality and Efficiency of Judicial Case Handling

Realizing the Case Separation Based on Complexity

For the purpose of building a green channel for enterprise-related cases to be effi-
ciently handled, cases must be accurately identified when entering the judicial
process. It is necessary to strengthen the construction of the data middle platform and
data governance. After the accurate definition of enterprise-related cases, enterprise-
related cases are tagged, and a data pool focusing on enterprise-related cases is estab-
lished by the addition of case category fields in the case handling system, automatic
identification, entry and correction of legal documents and other means to lay the
foundation for the following application in scenarios. Additionally, big data thinking
shall be applied in the whole establishment process of the standard case separation
system to ensure the person-case match,27 meaning that by comprehensive analyses
of factors including the types of a large number of completed enterprise-related cases
in the case handling system, the volume of files and documents, the number of people
involved, the case handling time, and the organization, procedures and methods of

27See Liu Guoyuan & Wenzhou, The Improvement of Procuratorial Case Distribution Mechanism
under the Condition of Big Data, People’s Procuratorial Semimonthly, No. 21, 2017.
6 The Use of AI in Law in Response to Major Epidemics 255

case handling, the rule for evaluating the complexity of enterprise-related cases is
determined, and the case distribution is more targeted to distribute enterprise-related
cases to the fast solution track or the normal solution track and to realize the careful
trial of complicated cases, speedy trial of simple cases and special trial of serious
cases.

Enhancing Intelligent Assistance in Handling Cases

In terms of unifying the case handling standards, the data of electronic reports,
review reports, legal documents and other documents of enterprise-related cases that
have been completed in the case handling system shall be extracted, integrated and
analyzed to form a knowledge map of the elements of crimes so that the judicial
standards can be unified and the problem of different judgments for similar cases
can be eliminated. Meanwhile, a standard evidence examination standard shall be
established, covering forms of evidence necessary for handling similar cases, the
legality of evidence collection, the way of evidence examination, the exclusion of
illegal evidence and other rules. The standard shall be embedded in a case handling
system for intelligent correlation review, evidence screening and verification, and risk
warning. With regard to improving case handling, big data technology can be used to
perform intelligent assistance for sentence suggestions, which compares and analyzes
a large number of sentencing recommendations of public prosecutors for tried cases
and court decisions, calculates the average ranges of sentencing recommendations
for similar cases based on algorithms to provide a predictive reference for sentencing
recommendations and final decisions for cases in progress. For improving the case
handling experience, the efficiency of case handling can be improved by intelligent
search and automatic push of laws and regulations, judicial interpretations, guiding
cases, similar cases, etc., which is realized by natural semantic recognition, big data
analysis, etc. and based on the combination of the features of cases in progress and
the precise portraits made by judicial personnel.

Improving Intelligent Case Management

The supervisory management of cases is a significant means to ensure quality and

efficient cases. As the judicial accountability system continues to be further imple-
mented, the supervisory management of cases is required to adapt to the new demands
of the reform and improve itself in terms of information technology and intelligence.
In the handling of enterprise-related cases, the application of big data in scenarios can
be strengthened to conduct specialized process supervision and quality assessment
and to promote the dynamic analysis and visualization of enterprise-related case
supervision data, which can provide a reference for business decision-making. In the
process supervision, the key nodes and risk points of the handling of enterprise-related
cases shall be untangled and summarized, such as listening to opinions, investiga-
tion and verification, disposal of involved money and materials, etc., which shall be
256 Y. Cui

included as supervisory rules with which the system automatically carry out patrols
so that the whole process of case handling is under dynamic supervision. In terms of
quality assessment, for problems in handling enterprise-related cases such as whether
there is no prosecution after arrest, inconsistent prosecution and sentencing affecting
the nature of cases or cross-range sentencing, the big data of completed cases shall be
analyzed and researched to conclude automatic identification rules for case quality
problems and to accurate screening and labeling and assessment guidelines of the
assessment system to improve the efficiency of supervisory management.

6.3.2 Deepening the Comprehensive Market Management

Strengthening the Data Connectivity and Sharing

For the purpose of fully upgrading the application efficiency of the 206 System,
which currently fails to be comprehensively connected with all political and legal
systems, it is necessary to upgrade the efficiency and scope of data collection and
minimize the negative impact of delayed data exchange and insufficient sharing on
case handling. In the long run, it is necessary to promote a standard and intelligent
information exchange and set up a unified comprehensive management platform for
administrative and judicial information so that the information systems of administra-
tive and judicial organs can intelligently and seamlessly connect with each other and
share data in real time. For the connection between administrative law enforcement
and criminal justice, based on the existing data sharing platform, collaboration and
external protection should continue to be strengthened, data from all parties should
be entered in a timely and accurate manner, and OCR technology should be used
to improve entry efficiency at the same time. In the promotion of joint development
and governance as well as the sharing of regional judicial big data, it is necessary
to improve big data exchange channels or establish a unified platform through the
collaborative mechanism for judicial organs in the Yangtze River Delta region and
promote standard and unified judicial case handling around the region to effectively
solve problems in the delta, such as inconsistent judicial standards and procedures for
handling enterprise-related cases and inconvenient electronic evidence acquisition.

Promoting AI-Assisted Supervision

When conducting legal supervision, procuratorial organs can collaborate with other
parties concerned to promote comprehensive market governance. Based on the collec-
tion and sharing of law enforcement data and judicial data, procuratorial organs can
sort out, conclude and summarize key points for supervision from frequent problems
in enterprise-related administrative law enforcement, criminal justice and civil and
administrative litigation cases, transform them into identification rules embedded
in the system to automatically screen case clues, make warnings and promptly
include clues into prosecutions, and initiate legal supervision work such as early
6 The Use of AI in Law in Response to Major Epidemics 257

intervention and advice on correcting illegal recommendations. For example, for the
problem of economic disputes being intervened in by means of criminal punishment,
which is strongly complained upon by enterprises, certain rules can be formulated to
compare the data correlation between administrative law enforcement and judicial
cases related to involved enterprises and comprehensively assess the risk of supervi-
sion. In the procuratorial work for public interest litigation, the data of administrative
law enforcement can be compared to promptly screen for clues in food and drug,
environment and resources, citizen information protection and other areas; data anal-
ysis and prediction can also be conducted for cases that have been handled, and for
fields and types prone to problems, clues of cases can be explored on platforms such
as the “connection between administrative law enforcement and criminal justice”.

Precise Supervision Advice

Judicial advice and procuratorial advice are effective ways for judicial organs to
serve economic and social development. Take the procuratorial work for example.
Currently, procuratorial organs tend to make procuratorial recommendations for
problems of the same kinds found when handling individual cases, and they tend
to make recommendations for individual cases so that procuratorial recommenda-
tions have not fully played the role of providing a reference for all similar cases
when one case has been handled. During the optimization of the doing-business
environment, judicial organs can focus on specialized areas, specific industries or
new business models, enhance the analysis and research on big data of similar cases,
accurately identify the problems in the production, operation and management of
enterprises and, by analyzing the causes, give more targeted and practical judicial
and procuratorial recommendations to help the enterprises involved solve the prob-
lems. Meanwhile, judicial organs can also sort out the data of similar cases that
frequently occurred within a period of time, analyze and research the characteristics,
patterns, geographical areas, crime trends of enterprise-related cases and use judi-
cial and procuratorial recommendations, business analysis reports and white papers
to suggest and encourage relevant departments to improve the regulatory mecha-
nism, promote legal administration and create a favorable environment for enterprise
development.

6.3.3 Upgrading Judicial Public Services

Improving the Judicial Cloud Service

To further improve the convenience and intelligence of judicial services and enhance
the user experience, it is necessary to further integrate the content of services, such
as the integration of the data of receiving complaints made by means of letters, visits,
the Internet and telephone calls to avoid the fragmentation of the service of receiving
and processing requests, and the connection with the data of case handling system
258 Y. Cui

to automatically inquire for the procedural information of enterprise-related cases,

etc. On the service platform, the supply of judicial services on administrative service
platforms such as Government Online-offline Platform shall be improved, and the
sufficient collection and sharing of public data of all departments on the platform
shall be promoted to identify judicial requests of enterprises and clues of cases from
the massive data and explore the provision of one-stop enterprise related services
including licensing, registration, law enforcement, justice, execution and others to
increase the satisfaction of enterprises with administrative and judicial services.

Promoting Corporate Compliance

As an ethical norm that enterprises must abide by when participating in fair competi-
tion in the market, corporate compliance is broadly focused around the world. As the
establishment of a corporate compliance system is on the agenda, how to effectively
use judicial functions to encourage enterprises to improve the compliance system
and provide judicial assistance for the development of enterprises is an important
issue in the current judicial work. Other than the warning for legal risk prevention
and control, it is necessary to conduct data collection and analyses with specific
themes on the judicial function performance or handling results related to corporate
compliance for the purpose of big data development, such as the application of the
leniency system for admitting guilt and accepting punishment, decisions on relative
nonprosecution and the issuance of judicial and procuratorial recommendation on
corporate crimes, the dynamic tracking reminder for the compliance of subsequent
production and operation of the enterprises involved in legal compliance, and boost a
good culture of corporate compliance. Additionally, based on judicial big data collec-
tion, judicial organs can explore and promote a blacklist system jointly established
with all parties concerned, which will be incorporated into the social credit system
to promote business integrity and fair competition.

Providing Distinctive Judicial Services

Take procuratorial work for example. Before the transformation of the department of
investigation and prevention of duty-related crime, procuratorial organs were used
to provide the public with the bribery crime archive inquiry service, which was
terminated upon the transformation. To better serve the construction of safe and
rule of law China, it is necessary to provide distinct judicial public services in the
fields related to people’s livelihood, public interests and business development with
the overall framework of the development of judicial big data. For instance, a case
distribution map can be drawn based on the data results of intelligent clustering.28
By extracting and analyzing data of enterprise-related cases of types such as property

28See Tao Jianping, The Value of Big Data in Procuratorial Work, People’s Procuratorial
Semimonthly, No. 10, 2018.
6 The Use of AI in Law in Response to Major Epidemics 259

misappropriation, selling fake goods and illegal fundraising that occur in a certain
period and using the geographic information system (GIS), crime hotspot maps are
formed to make visualized presentations or offer interactive inquiry services. The
maps present the change and trend of enterprise-related crimes in specific regions
in spatial form to offer alert and reminder services for police deployment, enter-
prise security, government regulation and investment decision-making, which will
boost the improvement of a multidimensional prevention and control mechanism and
effectively reduce enterprise-related crimes.29

29For example, the Jinshan District People’s Procuratorate in Shanghai took the Fengjing town
under its jurisdiction as a pilot area and mapped out crime hotspots to provide strong support for
improving comprehensive social governance capability.
Chapter 7
Expert Interviews on the Development
of Artificial Intelligence and the Rule
of Law

Yadong Cui

1 How Shall Legal Professionals Face an Era of Artificial

Intelligence?—Interview with Cui Yadong, Secretary
of the Leading Party Members Group and President
of Shanghai Law Society

Cui Yadong: Secretary of the Leading Party Members’ Group and President of
Shanghai Law Society, and Justice of the Second Rank. He once served as Director
of the Public Security Department of Anhui Province. In Guizhou Province, he was
a member of the Standing Committee of the Provincial Party Committee, Secretary
of the Political and Legal Affairs Committee and Director of the Public Security
Department, and President of the Guizhou Law Society. He is also the former Pres-
ident of Shanghai High People’s Court. His works include Artificial Intelligence
and Judicial Modernization, A Rule of Law Country, and Emergency Management
and Social Governance of Group Incidents. Cui has served as editor of a variety of
books, including Exploration and Practice of the Reform of the Judicial System in
Shanghai Courts, The Shanghai Sample of the Construction of the 12,368 Litigation
Service Platform, and Sunlight Illuminating the Return Road. His book Artificial
Intelligence and Judicial Modernization was designated by the National Office for
Classic Projects as the “2019 Classic China International Publishing Project” and
was translated into English and published by the German Springer Press.
The Blue Book: AI is changing the world profoundly. How does its development
impact the process of judicial modernization? You have said that embracing new
technology is the only way to judicial modernization? Why is that?
Cui Yadong: As a strategic technology leading the future, a new generation of AI is
thriving worldwide, injecting new momentum into economic and social development
and tremendously shifting the way people live and work. New technologies, such

Y. Cui (B)
Shanghai Law Society, Shanghai, China
e-mail: [email protected]

© Shanghai People’s Publishing House 2022 261

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2_7
262 Y. Cui

as AI in particular, have significantly contributed to the fight against the pandemic,

which makes people gain a deeper understanding of the capability and function of
AI in future development.
While changing the world, AI is also penetrating the judicial field.
How do legal professionals deal with the advent of the AI era? What is their
responsibility in this era? At least, I think, there are two points. First, legal profes-
sionals should embrace new technology, seize the opportunity, and actively promote
the deep integration of AI and rule of law. Judicature with the application of tech-
nology can better maintain social fairness and justice (AI + rule of law). Second,
legal professionals should vigorously enhance the establishment of AI and rule of law
systems, create AI and rule of law ecology, and promote, standardize and guarantee
the safe, reliable, controllable and sound development of AI (AI and rule of law) for
the betterment of humanity.
Legal professionals’ attitudes and responsibilities in the AI era proactively
embrace new technology and promote the deep integration of AI and justice.
In July 2017, General Secretary Xi Jinping noted that “efforts should be made
to follow the laws of judicature, combine the deepening of judicial systems reform
with the application of modern technology, and continuously improve and develop
socialist judicial system with Chinese characteristics” (instructions on promotion of
national judicial system reform session on July 2, 2017).
Marx once stated, “A science is truly developed only when it has reached the point
where it can successfully apply mathematics.”
Deng Xiaoping proposed, “Science and technology is the primary productive
force (1988).”
“Economics witnessed rapid development because of the introduction of mathe-
matical models and algorithms. In comparison, the science of law lags behind the lack
of modern technologies such as intelligent algorithms. The introduction of machine
deep learning, evidential models, algorithms, etc. in judicature leads to more scien-
tific and accurate judicial activities, and improvements in dealing with judicial cases”,
said Zhouqiang, President of the Supreme People’s Court.
Whoever has an advantage over AI will be able to gain the initiative in the
competition to win the future.
Legal professionals and scientists have been unremittingly pursuing the common
dream of judicial modernization by applying technology in the judicial field. The
realization of such a dream requires the unremitting efforts of generations.
We are in the era of AI. Legal professionals are supposed to be highly aware
of strategy, science and technology, as well as opportunity. We should firmly take
the rare historical chance of a new generation of AI, focus on advancing AI +
justice, AI + law-enforcement, AI + governance and so on, and adopt and apply
new technologies such as machine deep learning, evidential models, and algorithm
new technologies into judicature. These efforts are made to undertake more scientific
and accurate judicial activities, assist judicial case handling, improve the quality and
efficiency of case handling, and enhance judicial credibility. Only when justice is
combined with science and technology can it develop and progress better, become a
true science and better demonstrate the essence of fairness and justice.
7 Expert Interviews on the Development of Artificial Intelligence … 263

The only way to realize judicial modernization is to embrace new technology and
make full use of AI.
The Blue Book: You are praised as the chief designer of the “206 System” and
have organized its research, development and application. Would you like to talk
about it?
Cui Yadong: Chief designer is too much for me. The research and development
of the “software for trial-oriented litigation system reform” (referred to as the “206
System”) is a major task the Political and Legal Affairs Central Committee assigned
to Shanghai High People’s Court on February 6, 2017.
Early in 2014, Shanghai became the first pilot municipality among provinces and
cities to reform its judicial system. Then, in February 2017, Shanghai was respon-
sible for developing the “206 System”. Zhouqiang, President of the Supreme People’s
Court, remarked that judicial reform and information application are critical to devel-
oping judicature. Shanghai undertakes this arduous task to achieve breakthroughs in
judicial reform and IT application.
I think the Political and Legal Affairs Central Committee gave the task of
developing the “206 System” to Shanghai from the following backgrounds:
First, the background of correcting unjust, false and misjudged cases is in
accordance with the law.
Since the 18th CPC National Congress, against the background of advancing
law-based governance, the people’s courts have corrected 46 major unjust, false and
misjudged criminal cases, including cases involving Huugjilt and Nie Shubin, which
raised people’s confidence in judicial fairness (announced by the Supreme People’s
Court in February 2019). However, correction is far from sufficient, and we must be
determined to fundamentally prevent the occurrence of unjust, false and misjudged
cases.
Second, the background of promoting the trial-centered litigation reform.
How can the occurrence of unjust, false and misjudged cases be fundamentally
prevented?
In 2013, the third plenary session of the 18th CPC approved The Decision on
Major Issues Concerning Comprehensively Deepening Reforms by the CPC Central
Committee, which kicked off the judicial system reform. In 2014, the fourth plenary
session of the 18th CPC adopted the Resolution of the Central Committee of the
Communist Party of China on Major Issues Concerning Comprehensively Advancing
the Rule of Law, in which one of the reforms was to promote trial-centered litigation
reform. General Secretary Xi Jinping noted in his speech at the session that the goal
of promoting the trial-centered litigation reform is to encourage case handlers to
establish the concept that the cases handled must withstand test of judicial scrutiny,
ensure that the factual evidence of the case under investigation, examination and
prosecution can withstand test of judicial scrutiny, and ensure that the court trial
plays a decisive role in ascertaining facts, identifying evidence, protecting the right
of action and making a fair judgment. The reform helps case handlers have a stronger
sense of responsibility, helps realize substantive fairness in case handling through
procedural fairness in court trials, and helps effectively prevent the occurrence of
unjust, false and misjudged cases.
264 Y. Cui

The implementation of this reform will efficiently avoid such cases and ensure
that the innocent is not criminally prosecuted and that the guilty is justly punished.
It reflects the essence of the Chinese socialist rule of law and its progress. Promoting
the trial-centered litigation reform is not only the great historical background but
also the purpose and significance of developing the “206 System”.
Third, the development of new technologies such as AI provides great historical
opportunity and solid technological support for this purpose.
How to implement this reform?
Science and technology are powerful tools to solve problems. New technologies
such as AI provide great historical opportunity and solid technological support for
trial-centered litigation reform, for the prevention of unjust, false and misjudged
cases, and provide the path and means to achieve this goal, making it impossible.
As the president of Shanghai High People’s Court at that time, I had the privilege
to take part in developing the “206 System” (Now I am still in charge of it).
Over the past 3 years, under the leadership of the Political and Legal Affairs
Central Committee, Shanghai committee of the Communist Party of China, and
Shanghai Political and Legal Affairs Committee, with the support of the Supreme
People’s Court, the Supreme People’s Procuratorate, and relevant departments in
Shanghai Municipality, led by Shanghai High People’s Court, the departments of
public security, procuratorate, justice, and law-enforcement in Shanghai closely assist
each other and cooperate with iFlytek to firmly hold the chance of a-new-round tech-
nological innovation, and fully utilized modern technologies such as Internet, big
data, AI and blockchain to successfully develop Shanghai Intelligent Assistant Case
Handling System for Criminal Cases, a trial-centered reform software for litiga-
tion system (hereinafter referred to as the “206 System”). This is a good example
of integrating modern scientific and technological power with judicial personnel’s
creativity, forming an effect of scientific and technological rationality, judicial ratio-
nality and human rationality, creating a precedent for the deep application of AI in
the judicial field, creating a new way for reform, and being in a leading position at
home and abroad.
First, it advances the implementation of trial-centered litigation reform (boost
reform).
It is significant to prevent the occurrence of unjust, false and misjudged cases,
respect and protect human rights, ensure that innocent people are free from crim-
inal prosecution and that those convicted be justly punished, and promote the
implementation of trial-centered reform of litigation systems.
Second, it pushes forward innovative development of AI application in the judicial
field (technological revolution).
Modern technologies, including big data and AI, should be incorporated into crim-
inal litigation; a unified evidential standard should be established; such standards
should be embedded in digitized case handling procedures; examine, proofread and
check evidence; assist the work of case handlers; ensure that the facts and evidence
of cases subject to investigation and examining prosecution withstand legal inspec-
tion; and prevent the occurrence of unjust, false and misjudged cases. These are the
7 Expert Interviews on the Development of Artificial Intelligence … 265

breakthroughs of the deep application of AI from 0 to 1 in the judicial field, which

makes justice a real science.
Third, it helps to tackle the problem that has long hindered judicial fairness and
justice (crack the problem).
Currently, judicial system reform encounters numerous problems and difficulties
as it enters into a crucial period. General Secretary Xi Jinping pointed out that it was
no less difficult than the reform 40 years ago. How can judicial reform be deepened
and institutional barriers be removed? How can judicial quality be further raised,
unjust, false and misjudged cases be evaded, and judicial fairness be guaranteed?
How can we further elevate the openness and transparency of justice and enhance the
function and efficiency of judicial services? How can we strike a balance between
overwhelming cases and a limited number of case handlers and improve judicial
effectiveness? These issues need to be addressed by deepening reform (ideology and
institutional mechanisms) and applying technology. Practice has proven that science
and technology are powerful tools to promote reform and solve problems.
We shall apply modern science and technology such as artificial intelligence to
judicial activities so that the great power of modern technology and the wisdom of
judicial personnel can be better combined to form an effect of scientific and techno-
logical rationality, judicial rationality and human rationality, to maximally overcome
the limitations of judicial people’s understanding and subjective arbitrariness, mini-
mize errors in work, eliminates interference from interpersonal relationships, effec-
tively prevent the occurrence of unjust, false and misjudged cases, improve judicial
quality and efficiency, ensure fair justice, and enhance judicial credibility.
The Blue Book: In September 2018, the Shanghai Law Society held the High-
level Seminar on AI and Rule of Law at the World AI Conference and released the
Shanghai Initiative on Artificial Intelligence and the Construction of the Future Rule
of Law. At that time, the concept “AI and rule of law” was trail blazing around the
world. Can you share some views, based on judicial practices, on how to further grab
the chance and boost judicial modernization, to ensure social fairness and justice, as
well as provide legal protection for the implementation of national AI strategy?
Cui Yadong: On September 17, 2018, Chinese President Xi Jinping noted in
his congratulatory letter to the 2018 World AI Conference (Shanghai) that “a new
generation of AI is thriving worldwide, injecting new momentum into economic
and social development, and tremendously shifts the way people live and work. The
seizure of this opportunity for development to address new problems of AI in law,
safety, employment, moral ethics, social governance and so on requires deepening
cooperation and joint exploration by all countries. China wishes to work together
with other countries to accelerate development, protect safety, and share the fruit of
AI”.
AI is a double-edged sword. As it benefits mankind, it brings risks and challenges.
Focusing on AI’s rapid development, people are concerned with its potential hazards.
How can we cope with its potential risks and challenges?
The rule of law is the best way of governance and the most important condition
for development. AI development also needs rule of law as a guarantee.
266 Y. Cui

As I have mentioned, in the era of AI, it is a duty of legal professionals to vigorously

boost the construction of AI and rule of law systems, foster an environment for AI
and rule of law, adopt rule of law mentality and methodology, address risks and chal-
lenges, and lead, promote, standardize and guarantee the safe, reliable, controllable
and sound development of AI.
It is a new task in front of us to cope with the possible risks and challenges of AI
development. Shanghai academia of jurisprudence and legal profession have taken
an active part in responding and taking action, and achievements have been made.

1.1 Building AI and Rule of Law Hubs, Establishing AI

and Rule of Law Systems, Fostering an Environment
for AI and Rule of Law, and Enabling AI to Better
Empower Society

On June 20, 2017, the Standing Committee of the CPC Shanghai Municipal
Committee held a study session with the theme of “Development and Applica-
tion of International and Domestic Artificial Intelligence” at the Shanghai High
People’s Court, focusing on the development and application of international and
domestic artificial intelligence (over 300 members of the Standing Committee of
the Shanghai Municipal Committee, the Shanghai Municipal People’s Congress,
the Shanghai Municipal Government, the Shanghai Municipal Committee of the
Chinese People’s Political Consultative Conference and various municipal depart-
ments, districts, colleges and universities, and large enterprises attended the session).
In the session, I presented the effectiveness of applying AI in Shanghai courts. iFlytek
reported “the latest progress and typical applications of AI”. Comrade Han Zheng,
member of the Political Bureau of the CPC Central Committee and Secretary of
the Shanghai Municipal Committee, delivered a speech in which he proposed that
“Shanghai should strive to build a national hub for AI development.”
In September 2018, Comrade Li Qiang, a member of the Political Bureau of
the CPC Central Committee and Secretary of the Shanghai Municipal Committee,
stated in his address at the opening ceremony of the 2018 World Artificial Intelligence
Conference that “Shanghai will rely on the advantages of scientific and educational
resources, application scenarios, massive data and infrastructure, focus on innovative
curation, demonstrative application, system supply and talent gathering with a global
and future-oriented vision, and accelerate the construction of the ‘Shanghai Hub’ for
AI development” (referred to as the “four hubs”).
The Shanghai academia of jurisprudence and legal profession vigorously
responded and acted to serve the implementation of the national AI strategy and the
construction of the “four hubs” and proposed building a hub of AI and rule of law. It
makes clear that the construction of AI and rule of law hubs should be global-oriented,
forward-thinking, peace-oriented, people-oriented, safe, for innovative development,
sharing results, reliable and controllable, standardized and orderly. Combining AI
7 Expert Interviews on the Development of Artificial Intelligence … 267

and rule of law, as well as applying rule of law mentality and methodology to lead,
promote, standardize and guarantee the sound and sustainable development of AI,
can better benefit human beings.

1.2 Building an AI Legal Research and Exchange Platform

and Promoting AI Global Governance

It is an important duty of the Law Society to lead legal research with prosperity. The
Shanghai Law Society gathers the strengths of parties to actively build an exchange
platform for legal research on AI, to seize the high ground in the construction of
AI and rule of law system and the international discourse on system supply, and to
promote AI global governance.

1.2.1 Holding a High-Level Forum on “AI and Rule of Law”

and Building an International Exchange Platform for AI
and Rule of Law

High-level forums on “AI and Rule of Law” have been held at the World Artificial
Intelligence Conference for three consecutive years (including 2020). The theme
of the 2018 Forum is “Development, Application and Legal Protection” (over 300
people attended the forum). The theme of the 2019 Forum is “Co-Constructing the
Rule of Law for the Future, Sharing the AI Benefits” (more than 800 people, including
nearly 100 foreign scholars from 23 countries attended the forum). There were 11
specialists and scholars from China, the United States, the United Kingdom, Holland,
etc. who delivered keynote speeches successively, which exerted remarkable influ-
ence on the world. The theme of the 2020 Forum is “Rights, Obligations, and Legal
Practices Concerning AI”, and 10 experts from China, the United States, Sweden,
Japan and Korea will give speeches.

1.2.2 Releasing the Shanghai Initiative on Artificial Intelligence

and the Construction of the Future Rule of Law and Proposing
a New Concept of Rule of Law for AI

The Shanghai Initiative on Artificial Intelligence and the Construction of the Future
Rule of Law was issued at the 2018 High-Level Forum on AI and the Rule of Law.
This is the first initiative on AI and the rule of law at home and abroad, proposing 14
initiatives, including a conceptual framework for the construction of AI and the future
rule of law, a rule of law path to promote, standardize and guarantee AI development,
strengthening education, research and practice in the legal field of AI, and enhancing
international exchange and cooperation on the future rule of law of AI.
268 Y. Cui

1.2.3 Releasing the Blue Book of Artificial Intelligence and Rule of Law
in the World (2019) and Advancing AI Global Governance

The Blue Book of Artificial Intelligence and Rule of Law in the World (2019)
was released at the Rule of Law Forum at the 2019 World Artificial Intelligence
Conference. This is the world’s first book titled rule of law for artificial intelli-
gence. The Blue Book systematically examines the general situation in the field of
AI and rule of law, composes, summarizes, analyzes and makes relevant recom-
mendations by topic. Upon its publication, the Blue Book was listed as one of the
“Seven Actions” in the Three-Year Action Plan for Building the Artificial Intelligence
Highland and Constructing a First-class Innovation Ecology in Shanghai (issued
by Shanghai Municipal Commission of Economy and Informatization, Shanghai
Jingxin Zhi [2019] No. 707) and was incorporated as an essential part of Shanghai’s
AI development strategy.

1.2.4 Publishing Guidelines on Artificial Intelligence Safety and Rule

of Law (2019) and Outlining Strategies in Safety and Rule of Law

The Shanghai Law Society, together with the World AI Security High-end Dialog
2019, jointly issued the Guidelines on Artificial Intelligence Safety and Rule of Law
(2019). The guidelines make scientific predictions on the safety risks of AI develop-
ment from five aspects, including algorithm security, data safety, intellectual property
rights, employment and legal liability, and propose safety and rule of law strategies
to guard the security of AI development.

1.2.5 Issuing Youth Responsibility in the Era of Artificial Intelligence

and Advancing Legal Research on AI and the Growth of Young
Talent

During the Rule of Law Forum of the 2019 World Artificial Intelligence Conference,
Shanghai Law Society and Shanghai University of Political Science and Law collec-
tively organized the “Artificial Intelligence Rule of Law Youth Forum”, in which
nearly 300 university students from all over the world participated in exchange
activities and the initiative document Youth Responsibility in the Era of Artificial
Intelligence was released, which received wide attention from society.
7 Expert Interviews on the Development of Artificial Intelligence … 269

1.2.6 Establishing Artificial Intelligence Rule of Law Research

Association and Internet Rule of Law Research Association
of Shanghai Law Society

The Artificial Intelligence and Rule of Law Research Association assembles AI

interdisciplinary talent to contribute wisdom and strength to the development of AI
and rule of law.

1.3 Promoting Interdisciplinary Integration and Innovation

and Cultivating Interdisciplinary Talents of AI and Rule
of Law

Talent is the key of core competitiveness. A shortage of high-level talent in AI and

rule of law is a weakness in AI development.
Colleges and universities have unique advantages in talent cultivation and bear
a heavy responsibility for establishing interdisciplinary subjects of AI and law and
accelerating the cultivation of interdisciplinary talent in AI and rule of law. As early
as October 2018, General Secretary Xi Jinping already pointed out in his speech
at the ninth collective study of the Political Bureau of the Central Committee of
the Communist Party of China that “artificial intelligence has the characteristics of
multidisciplinary synthesis and high complexity. It is necessary to strengthen the
construction of talent teams, create a variety of forms of high-level talent training
platforms with greater determination and stronger measures, strengthen the training
of reserve talents, and provide more adequate talent support for the development
of science and technology and industry.” “To strengthen the efforts to train reserve
talents, institutions of higher learning should adjust their discipline settings and
curriculum, expand the scale of graduate students specializing in artificial intel-
ligence, and continuously train an army of talents for the development of a new
generation of artificial intelligence.”
Shanghai is in urgent need of high-level talent to build AI hubs and AI and rule of
law hubs. To this end, Shanghai Law Society, together with iFlytek, supported and
assisted Shanghai University of Political Science and Law to establish the “College of
Artificial Intelligence and Law” in May 2019, and enrolled 40 undergraduates and 10
postgraduates in Artificial Intelligence and Law in September, being the first AI law
school at home and abroad to take undergraduate in AI law (China’s two universities
including Southwest University of Political Science and Law have established school
of AI and law thus far, eight have set relevant majors, and thirty have founded related
research institutions), which fully reflects the reform spirit of Shanghai University
of Political Science and Law to carry out instructions of General Secretary Xi, and
to be bold, pioneering and innovative, and will cultivate a group of students who
understand both AI and law for the country. It will train a group of interdisciplinary
talents who can master both AI and law.
270 Y. Cui

Reform never ends, so does the development of science and technology and the
rule of law. Legal professionals have a heavy and glorious responsibility. By keeping
a close eye on the frontier of science and technology and conducting legal research,
we will strive to contribute “Shanghai Solution” and “China Solution” to developing
an AI and of law system, creating an AI and rule of law ecosystem, building an
“AI and rule of law hub”, and ensuring the safe, reliable, controllable and healthy
development of AI.

2 Reflections on the Application of AI to the Operating

Mechanism of Judicial Power from the Perspective
of Procedural Justice—Interview with Ye Qing, President
of East China University of Political Science and Law

Ye Qing: Currently the Deputy Secretary of the Party Committee and President of
East China University of Political Science and Law. He is a member of the 15th
Standing Committee of the Shanghai Municipal People’s Congress, Vice President
of China Criminal Procedure Law Research Institute, Vice President of Shanghai
Law Society, and Vice Chairman of Shanghai Federation of Social Science Asso-
ciations. His research mainly focuses on criminal procedure law, evidence law, and
Chinese and foreign judicial systems. Ye Qing has undertaken a number of major,
general and commissioned projects for the National Social Science Fund of China
and has participated in nearly 20 provincial- and ministerial-level projects (including
commissioned projects); he has published more than 130 papers on core journals of
law and jurisprudence; books authored by him include Research on Evidence in Crim-
inal Litigation, Research on Litigated Criminal Pretrial Procedures, and Evidence
Law: Problems and Explanations. Additionally, he has edited and coedited over 30
books and textbooks.
Currently, there is an urgent need to sort out and regulate the operation mechanism
of judicial power from the perspective of procedural justice.
The Blue Book: Hi, President Ye, as an expert in criminal procedure law, how
do you view new technologies such as AI from the perspective of procedural justice,
and adjustment and reform of judicial power operation mechanism triggered by these
new technologies?
Ye Qing: In recent years, AI has received unprecedented attention and promotion
at the national strategic level,1 especially when the application of modern technolo-
gies such as mobile Internet, informatization and big data is on the rise. Judicial

1 Since 2017, AI has been included in the Government Work Report for three consecutive years,
referring to People’s Daily (Overseas Edition), March 11, 2019, p. 11. The State Council of China
issued The Development Plan of the New Generation Artificial Intelligence in July 2017, focusing
on accelerating the deep integration of AI and economy, social and national defense, and regarding
the promotion of scientific and technological innovation capacity of the new generation artificial
intelligence, to develop an intelligent economy and build a smart society.
7 Expert Interviews on the Development of Artificial Intelligence … 271

organs have also taken the construction of smart courts as an important path to
modernize the trial system and trial capacity.2 In this context, courts around China
have made extensive explorations in big data platform establishment, AI assistance
technologies, and “Internet+” litigation models.3 Despite the fact that many tech-
nological innovations are still at the stage of weak AI, or in other words, the deep
application of modern information technology, it should be noted that these reforms
have subtly affected and changed the operation mechanism of traditional judicial
power, and there is an urgent need to sort out and regulate the operation mechanism
of judicial power from the perspective of procedural justice.
The present legal research mainly concentrates on the impact of AI on the legal
system, legal qualification and subject matter of AI, as well as AI in substantive
criminal law and economic law.4 Academic research examining such development
trends from the perspective of procedural justice is not common. Popularity and
scarcity of research in substantive and procedural justice, respectively, will definitely
hamper the deepening reform of the judicial power operation mechanism and the
construction of a modern trial system that conforms to basic principles of procedural
justice, which will eventually lead to procedural justice challenges in judicial practice.
For instance, a court in Wisconsin employed “Correctional Offender Management
Profiling for Alternative Sanctions” in sentencing and identified a defendant with a
high possibility of recidivism, who then appealed to the Wisconsin Supreme Court
on the grounds that the judge had violated due process doctrine and that it was not
objective and neutral.5 This case reminds us of the ongoing smart justice that we
should avoid procedural tools and nihilism and apply the value of procedural justice
to evaluate, regulate and guide the adjustment and reform of the judicial power
operation mechanism triggered by technology.
Speedy trial: how to deliver justice timely?
The Blue Book: Justice delayed is justice denied. This is also reflected in the Criminal
Procedure Law of the People’s Republic of China, which stipulates “A People’s Court
shall pronounce judgment on a case of public prosecution within two months or,
three months at the latest, after accepting it”. Speedy trials are also a basic litigation
principle recognized by the two major legal systems. However, in judicial practice,
commencement of a speedy trial is subject to various factors, both internal and
external. The trial schedule is influenced by external and objective elements. The

2 See SPC president calls for strengthening application of block chain, AI by Sun Hang published
on People’s Court Daily, November 10, 2019, p. 1.
3 For example, four open platforms of trial procedures, judgments, court hearings, and enforcement

information, and Legal Information Platform, following with Shanghai Intelligent Assistant Case
Handling System for Criminal Cases, “AI Judge” system in Beijing, Guiyang political and legal big
data project, Suzhou intelligent trial model, Hangzhou Internet Court, etc.
4 See Rethinking Chines Law and Artificial Intelligence by Zuo Weimin published on Global Law

Review, Vol. 2, 2019.

5 See Rethinking Chines Law and Artificial Intelligence by Zuo Weimin published on Global Law

Review, Vol. 2, 2019; Artificial Intelligence in the U.S. Legal System: Challenges and Concerns by
Benjamin L. Liebman published on China Law Review, Vol. 2, 2018.
272 Y. Cui

schedule of defense lawyers (especially when multiple defendants and more lawyers
are involved, it’s hard to fix the time of court session due to business travels of
lawyers), witnesses, work arrangements, etc., are illustrations of why the opening of
a court session cannot be scheduled timely. Under the influence of COVID-19, many
cases cannot be heard in traditional court mode. The application of the suspension
procedure will lead to prolonged delays and replacement of sentences by custody.
What issues do you think need to be noted in online trials?
Ye Qing: Online trial is nothing novel. In September 2018, the Supreme People’s
Court promulgated the Provisions of the Supreme People’s Court on Several Issues
Concerning the Hearing of Cases in Internet Courts. However, the current prevention
and control of the pandemic has highlighted the advantages of hearing cases via the
Internet. From the onset of the pandemic to February 19, 2020, the Supreme People’s
Court, Shandong High People’s Court, Shanghai High People’s Court, Shanghai No.
1 Intermediate People’s Court, Intermediate People’s Courts in Changzhou, Wuxi,
Nantong and Yingkou, and District People’s Court in Hami Prefecture, Xinjiang
Province, successively issued provisions on “online trials”.
I believe we need to pay attention to four issues in online trials. 1. Defendant’s right
to speedy trial and right to refuse online trial. 2. The online trial network style should
conform to the structure of litigation, such as the position of the judge’s video window,
and the position of the prosecution and defense looks like an isosceles triangle to
reflect the judge’s centrality and impartiality. 3. The scenario and dress code of online
trial. A judge and a prosecutor should abide by standards of professional appearance.
A national emblem should be visible above a judge’s head, and a prosecutor’s emblem
should be visible above a prosecutor’s head. 4. When it is inappropriate to hold online
trials and there are objections to material and documentary evidence, it is necessary
to confirm the defendant’s identity (best evidence rule).
The Blue Book: Elements that affect a speedy trial are often found in difficult
and complex cases. What role do you think AI can play in pretrial case reading, such
as carding and analyzing evidential arguments and preparing trial outlines?
Ye Qing: To improve the judge’s efficiency of reading files, it is necessary to auto-
matically review whether the form of evidence meets the requirements of competent
evidence, automatically screen the existing evidence to ascertain facts and contra-
dictions, and analyze competent evidence and evidence chains according to the
legal requirements of competent evidence and basic specifications of evidence in
similar cases by AI. In view of China’s transferring the entire case to the prosecution
rather than simply relying on the indictment, judges need to read the case to sort
out and grasp competent evidence, facts that can be proven by existing evidence,
and contentious points in evidence so that they can better grasp those points in trial
and hear the opinions of the prosecution and defense based on relevant evidence. In
summary, and speedy trial cases, as the court does not generally conduct court inves-
tigations or the evidence is merely summarized and cross-examined, if the defense
fails to propose effective defense opinions based on evidence and facts, the court
needs to request the prosecution to provide additional evidence or investigate and
verify the facts on its own to ensure that the case evidence satisfies the requirements
of competent evidence and standards of proof. If the judge finds that the defendant
7 Expert Interviews on the Development of Artificial Intelligence … 273

may be a recidivist, it is necessary to demand additional investigation through case

reading because the public prosecution failed to retrieve relevant records of previous
convictions.
The Blue Book: As we can see, the National People’s Congress decided in 2014
to launch a national pilot program for speedy adjudication of criminal cases in
Beijing, Tianjin, and Shanghai, and the latest Criminal Procedure Law of the People’s
Republic of China revised in 2018 has prescribed speedy adjudication procedures.
In accordance with Article 222 Paragraph 1 of the Law, “Speedy adjudication proce-
dure may be applied to cases, which shall be tried by a single judge alone, if the case
is under jurisdiction of Primary People’s Courts, the defendants may be lawfully
sentenced to fixed-term imprisonment of no more than three years, the facts of a case
are clear, the evidence is reliable and sufficient, and the defendant pleads guilty and
agrees to apply speedy trial procedure.” At present, it is believed that opening a court
session is a principle, while absence of a court session is an exception. However, in
the case of speedy adjudication, as the facts of a case are clear, the evidence is reliable
and sufficient, and the defendant pleads guilty, the trial can be held in writing without
a court session. What do you think of this?
Ye Qing: First, the judge can investigate the voluntariness, knowledge and legality
of the defendant’s and the defense’s guilty plea outside the courtroom without having
to confirm it through a ten-minute trial. Second, strictly limit the application scope
of verdict without a court session to the following scenarios: where the defendant
pleads guilty and accepts punishment; where speedy adjudication process is applied;
where the defendant may be exempted from criminal punishment, punished with fines
exclusively, be sentenced to public surveillance, criminal detention, and imprison-
ment with a suspension of execution. Such cases are often minor, in which the facts
are simple and clear, the defendant is not subjectively malicious, and the judgment
is enough to achieve the purpose of punishment and education. Third, a mandatory
defense system should be introduced to ensure that criminal suspects have access to
defense lawyers in prosecution and trial. Fourth, the defendant should be given the
right to waive the trial.
Public trial: How can the change from formal openness to substantial openness
be realized?
The Blue Book: You think the current judicial openness is limited to its form and
result, but not yet to an extent of substantive openness. What can be done with regard
to public reasoning of documents?
Ye Qing: First, increase the weight of excellent judgment documents in the assess-
ment and reward of judges. Second, the court assessment mechanism should be
improved, and the mechanical ranking of reversed judgment, remand rate and other
indicators should be abolished. Third, strictly define the circumstances and proce-
dures for judges to be held accountable, clarify the boundaries among judges’ perver-
sion of justice and misconduct, serious faults and ordinary negligence, and set up
a neutral provincial body to determine judges’ responsibility and procedures for
liability and remedy. Finally, standardize the trial report sample, explicitly prohibit
recording fact ascertainment, evidence analysis, and legal application in trial reports,
274 Y. Cui

and clearly prescribe that a trial report should only be used to record issues relevant
to cases that are not suitable to be reflected in the judgment.
The Blue Book: As information dissemination is rapid and extensive in the
Internet era, it is necessary to regulate judicial openness. In your opinion, under
the current conditions, how can we establish a standardized system to disclose the
information according to the law and the system?
Ye Qing: I think, on the one hand, lift restrictions on live trial. First, attention
should be given to protecting the identity of defendants and witnesses. Second, for
cases that may require witnesses and experts to testify in court, the online opening of
court sessions should be avoided to ensure the accuracy and authenticity of witness
testimony.
On the other hand, remove constraints on access to online judgments. A defen-
dant’s identity should be concealed in judgments of misdemeanor cases to help
the defendant reintegrate into society. Since the abolition of re-education through
labor and criminalization of drunk driving in particular, the court has received a
large number of simple and minor criminal cases, while criminal antecedents are
detrimental for defendants to return to the community.
How do AI systems serve lawyers?
The Blue Book: It is evident that departments of public security, prosecution and
law enforcement are highly enthusiastic about building the AI system, while the
defense is less involved. With respect to intelligent functions of similar cases posting
and sentencing reference that affect judges’ conviction and sentence, the prosecution
and defense may not be able to participate in the judge’ evidence determination at
his/her discretion, which violates the doctrine of procedural participation, because the
algorithm and database are closed in existing intelligent system. It’s not fair especially
for the defense. What do you think should be done to solve these problems?
Ye Qing: Due to process doctrine, a fair trial derives from an adversary system
in which two parties confront each other equally. The equality of such confrontation
originates from principles of natural justice. Therefore, the prerequisite of the adver-
sary system is to be equal. The Criminal Procedure Law of the People’s Republic
of China (1996) incorporated reasonable elements of partyism, and the latest Law
further strengthens legal protection for duty lawyers and defense lawyers.
I suggest that, first, build a platform shared by public security, prosecution, and
law enforcement departments, and lawyers to handle cases, on which they can read
files online, go through entrustment formalities, and submit defense opinions and
relevant application materials. Currently, when criminal cases in Shanghai are heard
in court, the defense lawyer of a criminal case can submit materials to the court elec-
tronically on the lawyer service platform of the Shanghai Court or the Mobile Micro
Court. Meanwhile, the internal information of each unit should be well contained
and protected. As far as I know, the 206 System in Shanghai has created a function
for lawyers to read files online. In the function module, a lawyer applies to court to
read files online. After the judge’s examination and confirmation, the case files are
exported to the Internet, and the lawyer reads the files on the Internet.
7 Expert Interviews on the Development of Artificial Intelligence … 275

Second, in an effort to prevent judges from handling cases mechanically by relying

on intelligent systems, the current Shanghai 206 System is established on the basis
of an online library of legal documents of cases that have been rendered effective
judgments, and the prosecution and defense both have access to functions such as
sentencing reference and risk assessment. The two parties can determine the short-
comings of the system model, data, and algorithm and put forward specific sugges-
tions, thus substantially participating in the mental process of a judge’s case handling.
Trials and sentencing must be determined in a court session, and they should not be
made based on assistive case handling references, such as similar cases posting and
sentencing references.
Litigation Economy: How does AI make judicial products more efficient?
The Blue Book: In the background of informatization, electronic documents are
not included in papers of criminal judicial practice, and electronic delivery is not
included in the delivery method. What is your opinion?
Ye Qing: First, I think that along with the development trend of a paper-free
society, the technologies of electronic file generation, electronic signature, and indi-
vidual identification are becoming increasingly mature. Article 15, Paragraph 3 of the
Provisions of the Supreme People’s Court on Several Issues Concerning the Hearing
of Cases in Internet Courts provides that an Internet court may electronically serve
judgments after informing the parties of their rights and obligations and obtaining
their consent. Electronic delivery of legal documents should be gradually promoted
for the convenience of parties and counsels in signing. Second, courts should be
empowered to use systems such as facial identification to facilitate online examina-
tion of party identity and to ensure accuracy. In addition, a nationwide working plat-
form should be created for departments of public security, prosecution, law enforce-
ment and judicature to work together to realize electronic service of legal documents.
For example, when a pretrial investigation is entrusted to a community correction
agency elsewhere, the entrustment letter can be served directly via the system.
Traditionally, cost-effective litigation means that specialized state agencies use
as few human, financial and material resources as possible to accomplish the tasks
of criminal proceedings. Concepts such as cost-effectiveness and efficiency have
been discussed in the context of reducing the burden on judiciaries. In fact, both the
cost of operating judiciary and the means by which parties expect to obtain judicial
products have to be taken into consideration. Over the years, countries with two
major legal systems have explored procedural diversion, including the plea bargain
system instituted in the United States since the 1960s; stipulations that the Magistrates
Court in the United Kingdom may render a judgment in absentia if the possible
imprisonment is less than three months; penalty order procedure in Germany; Italian
summary trial procedure without a formal adversary hearing, etc. We have drawn
these experiences in the last round of judicial reform.
276 Y. Cui

3 Explorations of Cultivating Interdisciplinary Talents

of Artificial Intelligence and Law by Shanghai University
of Political Science and Law—Interview with Liu
Xiaohong, President of Shanghai University of Political
Science and Law

Liu Xiaohong: Currently the Deputy Secretary of the Party Committee and President
of the Shanghai University of Political Science and Law. She is a deputy to the 15th
Shanghai National People’s Congress and Vice Chairman of the China Society of
Private International Law and Shanghai Law Society. She once served as the Vice
Chairman of the International Chamber of Commerce Commission on Arbitration
and Diversified Dispute Resolution and the Director of International Exchange Center
and Vice President of East China University of Political Science and Law. She was
awarded the honorary title of “the Fourth Shanghai Outstanding Young and Middle-
aged Jurist” (2009).
The establishment of the School of Artificial Intelligence and Law serves the
national strategy and local development
The Blue Book: As the first university in China to enroll undergraduates majoring
in “Artificial Intelligence and Law”, why does SHUPL focus on this new field? Can
you briefly explain the purpose and process of establishing the School of Artificial
Intelligence and Law?
Liu Xiaohong: Since its foundation in 1984, SHUPL has been adhering to “Being
Based on Political Science and Law, Serving Shanghai, Facing the Country and
Having the Whole World in View”. The university has a close connection with this
industry. It has formed distinctive political and legal characteristics by meeting the
needs of the political and legal industry in academic research and talent cultivation.
On September 13, 2013, President Xi Jinping proclaimed in the Bishkek Summit
of Shanghai Cooperation Organization that China will set up the “China-SCO Inter-
national Judicial Exchange and Cooperation Training Base” at SHUPL and is willing
to train judicial personnel for other member countries. Relying on the China-SCO
base, SHUPL has accelerated its special development and focuses on national secu-
rity, SCO rule of law, global security governance, legal protection of the “Belt and
Road Initiative”, and so on. The university has endeavored to build special disci-
plinary clusters such as rule of law guarantee of urban governance, judicature of the
“Belt and Road Initiative”, and research on the “Belt and Road Initiative” security
and anti-terrorism. The university has also strengthened international talent training,
high-end foreign training, think tank research and national and global academic
exchanges, which plays a unique role in serving national strategies and the “Belt and
Road Initiative”.
With the advent of the information age, AI has been regarded as a national strategy.
On October 31, 2018, General Secretary Xi Jinping emphasized during a group
study of the CPC Central Committee Political Bureau on the current situation and
trend of AI development that “AI is an important driving force of a new wave of
7 Expert Interviews on the Development of Artificial Intelligence … 277

technological revolution and industrial change, and accelerating the development of

a-new-generation AI is a strategic issue that concerns whether China can seize the
opportunity of the new-round scientific and technological revolution and industrial
change.” In 2017, the State Council issued the New Generation of Artificial Intelli-
gence Development Plan. In April 2018, the Ministry of Education introduced the AI
Innovation Action Plan in Institutions of Higher Learning, proposing the disciplinary
and professional construction of “Artificial Intelligence + X”. Shanghai has proposed
speeding up the building of a “Shanghai Hub” for AI development by focusing on the
source of innovation, application demonstration, formulating policies and attracting
talent.
As an application-oriented university with distinctive political and legal char-
acteristics, SHUPL actively serves national strategies and Shanghai’s develop-
ment, closely aligning with industrial needs and the innovative fruit of integrating
Shanghai’s judicial system reform and AI—trial-centered litigation reform (referred
to as the “206 System” in Shanghai). Relying on the China-SCO base, the univer-
sity explores the interdisciplinary development of AI + law and the cultivation of
interdisciplinary talent. In May 2019, supported by Shanghai Law Society, iFlytek,
etc., SHUPL integrated the resources of the whole university and actively strived
for external resources and took the lead in establishing the Law School of Artificial
Intelligence in Shanghai. In the same year, the school enrolled the first batch of 40
undergraduates majoring in AI and law, which has attracted great attention from
society.
The SHUPL School of Artificial Intelligence and Law has grown rapidly since its
establishment and has seen initial results and positive impacts in academic forums
and teaching reforms. On August 30, 2019, the School, together with Shanghai Law
Society, cohosted the first “Youth Rule of Law Forum of World Artificial Intelli-
gence Conference”, which built an international platform for youth around the world
to show their talents and shoulder their responsibility. The forum is now perma-
nent, held annually in parallel to the World Artificial Intelligence Conference. The
“Artificial Intelligence + Law” talent training model was included in the Law Talent
Training Program of Shanghai Municipal Education Commission, and the “Project
on Constructing Discourse System of Systems, Rules, Standards and Procedures in
the Era of Artificial Intelligence” was incorporated in the Shanghai Municipal Think
Tank Construction Project. “Artificial Intelligence and Law” and “Programming for
Data Analysis” were selected for the teaching reform projects of colleges and univer-
sities in Shanghai. In addition, the School of Artificial Intelligence and Law, as the
secretariat of the AI and Rule of Law Research Institute of Shanghai Law Society,
also seeks to assist and serve AI and rule of law researchers in Shanghai and across
China, which has been highly recognized by practical and legal communities.
The Blue Book: The establishment of AI and Law School has drawn much atten-
tion from the community, and the first undergraduate enrollment majoring in AI and
Law has received undiluted enthusiasm from students and their parents. Why is the
School so popular?
Liu Xiaohong: I think there are several reasons.
278 Y. Cui

First, AI itself is a hot spot in today’s society. AI has penetrated various fields,
with high technology content and many blank spots. Society urgently needs inter-
disciplinary talents of computer technology and law. The school enrolled students
with math and science backgrounds to cultivate interdisciplinary and applied talents
of art and science who can adapt to the development needs of AI. That is why it is
popular among candidates and parents.
Second, it serves the national and Shanghai development strategies. SHUPL takes
the lead in establishing the School of Artificial Intelligence and Law in Shanghai
and enrolls candidates in undergraduate programs of AI and law, which meets the
needs of national and Shanghai scientific and technological development strategies.
Situated in Shanghai, the university combines its talent cultivation with Shanghai’s
advantages in location, technology, and rule of law, actively explores new directions
and ways of talent cultivation, and takes the initiative to develop Shanghai’s higher
education, which is bound to attract wide attention from all walks of life.
Third, SHUPL has gained strong support from the industry and enterprises. The
establishment of the School of Artificial Intelligence and Law was guided by the
leaders of Shanghai Law Society and supported by iFlytek, a well-known high-
tech company in the field of AI. Therefore, the university is not running the school
alone but together with the industry and enterprises. From the preparation to the
establishment and to its future development, we have cooperated and will continue
to cooperate with the industry and enterprises in all aspects.
We are conducting groundbreaking exploration in teaching materials and
faculty of AI and law with an open mind
The Blue Book: As a brand new subject, there is no special textbook on AI and law in
China or even in the world. What is SHUPL’s solution to the shortage of textbooks?
Liu Xiaohong: AI and law is an interdisciplinary subject. To enrich professional
teaching materials of AI and law, SHUPL has set up an editorial board of profes-
sional teaching materials of the AI and Law, which is mainly composed of renowned
experts from Shanghai Law Society, iFlytek, Shanghai University of Political Science
and Law, Shanghai political and legal departments, and domestic and international
jurisprudence and legal academia. It has drawn up plans for compiling a series of
teaching materials. Based on the significant achievements of Shanghai’s judicial
system reform in AI jurisprudence research, the sufficient resources of iFlytek and
the important platform of the Rule of Law Forum of the World AI Conference, it has
invited well-known experts from jurisprudence and legal academia to work together
to compile special textbooks for AI and law to contribute to application-oriented
talent training. The “206 System” in Shanghai has provided important experience
and information for these textbooks. The task of compiling and planning series
textbooks is divided into three stages: short-term, medium-term and long-term.
The short-term compilation task is to prepare for courses that will be offered in
September 2020’s teaching plan, including Introduction to Artificial Intelligence and
Law, Principles of Artificial Intelligence, Python Data Analysis, Legal Application
of Artificial Intelligence, Artificial Intelligence and Law, Intelligently Assisted Case
Handling Protocol, and Typical Case Analysis of Artificial Intelligence. The midterm
7 Expert Interviews on the Development of Artificial Intelligence … 279

compilation task refers to teaching materials required to complete tasks of the third
and fourth school year in the teaching plan, such as Digital Evidence and Blockchain,
Legal Big Data Analysis, Knowledge Discovery and Data Mining, Data Statistics
and Analysis, Introduction to Network and Information Security, Introduction to
Artificial Intelligence Ethics, and Legal Management Information System. The long-
term compilation task is about teaching materials that support AI and law major and
discipline and that also includes some monographs used for teaching.
The Blue Book: School of Artificial Intelligence and Law is a newly established
school. What is your plan for its faculty development?
Liu Xiaohong: Upon the foundation of artificial intelligence and law school, it
mainstreamed the power of the whole faculty, which embodied its concept of training
interdisciplinary talents of computer technology and legal expertise. At present, the
school is vigorously introducing talent in both “full-time” and “flexible” ways and
intends to bring in 1–2 pacesetters in AI and law research, 5–10 “AI + law” cross-
disciplinary professors, and “flexibly” attract 1–3 famous experts from home and
abroad within 3 years. Currently, there is a shortage of teachers in AI and law in China,
and we welcome talents from home and abroad who are devoted to research and
teaching in AI and law to join us, so that we can build the Artificial Intelligence and
Law School together and promote the professional development and talent training
of AI and law. We also hope that we can cultivate and train an excellent academic
team and faculty in the field of AI and law on this platform and play a leading and
exemplary role in this field.
Meanwhile, the School of Artificial Intelligence and Law proactively seeks
resources from society, establishes a council and an expert committee, and hires a
panel of practical experts from the industry and enterprises as part-time lecturers and
professional internship supervisors to teach practical courses and instruct students
on professional internships.
Blue Book: In 2019, 40 undergraduates and 10 graduates registered in artificial
intelligence and law school. What are the school’s training programs and future
orientations?
Liu Xiaohong: In 2020, the school will continue to keep the current enrollment
scale and further explore a path to cultivate Ph.D. and postdoctoral students, straining
to form an entire talent training system from undergraduate, graduate, Ph.D., and then
postdoctoral students in three years.
In the next 3–5 years, SHUPL strives to become a leading university in subject
construction, discipline construction, talent cultivation and scientific research of AI
and law, formulate professional and discipline construction standards in line with
its own characteristics, and offer master’s and doctoral degree programs. At the
same time, with the support of Shanghai Law Society, iFlytek and political and legal
departments in Shanghai, the school promotes openness and innovation, builds a
talent training system with deep integration of university, industry and enterprises
for collaborative education, endeavors to make itself an influential school with special
features nationwide, and rolls out applicable and replicable experience and models
to other AI and law schools, and the development of AI and law discipline in China.
280 Y. Cui

We shall strive to cultivate interdisplinary talent to serve national strategies

The Blue Book: What kind of talents do you think will be in demand in the future
as AI and rule of law continue to integrate?
Liu Xiaohong: The Artificial Intelligence and Law School takes “serving national
strategies and insisting on smart education” as its development principles and “intel-
ligent research and practical education” as its development goal. The school is
committed to nurturing talents with a combination of “AI + law” knowledge and
skills, capable of working in national judicial organs, administrative departments,
enterprises and public sectors, scientific research institutions and so on, who engage
in the research and practice of legal application, laws and regulations, ethical norms
and policy systems of AI. Specifically, two types of talent are included. First, culti-
vating AI legal talents who can adapt to legal needs in the era of AI, including those
who provide legal services for the legislative, enforcement, judicial and compliance
needs of AI. Second, legal AI talents who can apply AI in law, including training
talents for the application of AI in legislation, law enforcement, justice and law
abiding, etc. We expect to train a group of AI experts with law expertise and legal
experts with AI knowledge for society through continuous efforts.
The Blue Book: After listening to your introduction, I am aware that AI is indis-
pensable in the development of science and technology, the rule of law, and higher
education and will increasingly play a greater role. The School of Artificial Intelli-
gence and Law of Shanghai University of Political Science and Law has also shown
its unique value.
Liu Xiaohong: Now we are in the AI era, an era of smart education, also an era of
promoting the development of legal science as new liberal arts. The establishment of
artificial intelligence and law school and the cultivation of AI + law interdisciplinary
talent is precisely an attempt at our higher education to respond to the development
of times, actively link with the needs of the national strategy, promote the reform
of talent training modes and explore the construction of new arts. We believe that
with the support of Shanghai’s political and legal industries and relevant AI high-
tech enterprises, artificial intelligence and law school will develop rapidly, forming
distinctive features and accumulating valuable experience in institutional innovation,
academic research and talent cultivation and contributing SHUPL’s wisdom and
strength to the country, society and the times.

4 Keeping Pace and Taking Initiative to Make AI Better

Serve the Development of Rule of Law—Interview
with Liu Qingfeng, Chairman of iFlytek

Liu Qingfeng: Chairman of iFlytek. He was admitted to the University of Science

and Technology of China in 1990. He founded iFlytek Co. Ltd. in 1999 and has
served as the president since then and as the chairman since April 2009.
7 Expert Interviews on the Development of Artificial Intelligence … 281

From perception to cognition: accelerating AI application in the judicial field

The Blue Book: As an expert and explorer in AI, how do you view the relationship
between AI and rule of law in general?
Liu Qingfeng: We can see that AI has been elevated to a national strategy and that
AI is an important driving force of a new-round technological revolution and indus-
trial change. It has been widely adopted in all aspects of human society, profoundly
affects and changes the world, and penetrates the judicial field. As shown in the prac-
tice and results of applying AI in the judicial field, the deep integration of modern
technologies, including AI and judicature technologies, is the only way to achieve
judicial modernization. Such a combination will play a major role in building civi-
lized judiciaries, making judicial progress, and improving and developing socialist
judicial systems with Chinese characteristics.
As AI continues to integrate deeply and grow rapidly in all areas of society, it
will have a huge effect on the legal system, economic development, social stability,
and even global governance. Generally, AI brings both challenges and opportunities
to the rule of law. It not only generates various new problems in typical scenarios
of practicing existing laws with regard to the intellectual property system and legal
liability determination and attribution but also further raises concern on basic issues
of the rule of law, such as the concept of a legal subject, and worries about future
AI destruction of the present human rule of law order. Therefore, to take profit and
avoid loss, how to control AI development has gradually become breaking news for
people from all walks of life.
The Blue Book: There is an idea that AI algorithms will be able to express the
law itself and even to legislate laws, enforce laws and administer justice. Do you
consider it possible?
Liu Qingfeng: At the moment, I do not think it is possible. German sociologist
Max Weber had a vision a hundred years ago: “Contemporary judges are vending
machines that are inputted with pleadings and litigation fees and output judgments
and reasons copied from codes.” His concept of formal rationality is reflected in judi-
cature as “syllogism”. It regards legal basis as a major premise, legal facts collected
from life as a minor premise, and based on such facts, a judge seeks a legal basis
and then makes legal decisions. This process seems suitable for AI because a judge’s
free evaluation of evidence can be recognized as an algorithm that has no connection
with human feelings, no prejudice or character, which is almost perfect; moreover,
it can deal with a human judge’s experience and intuition—AI is able to do transfer
learning.
However, things are not that simple. Law deals with concrete objects in human
society, which is full of various interest considerations and unknown matters that may
arise at any time. Machines can imitate only what can be clearly conveyed in explicit
formalized language and are less likely to replace purely rationalized disciplines and
vice versa. The most typical disciplines are literature and mathematics. For example,
to improve lagged and stereotypical law, the law shall grant judges discretionary
power that “ensures the verdict is in line with the wisdom of life”, which is now
beyond AI’s ability.
282 Y. Cui

The Blue Book: Which stage do you think AI application in the judicial field is
at now?
Liu Qingfeng: I think AI application undergoes rapid growth, from initial intelli-
gent perception, gradually rising to cognitive intelligence. A variety of AI application
products have sprung up in public security departments, procuratorates, courts and
judicial administrations, and judicial units at all levels are constantly exploring deep
AI applications based on the foundation of big data centers, which places an increas-
ingly higher demand for AI technology. Thus, there will be more possibilities for the
application of AI technology in the judicial field in the future.
Improving quality and efficiency: changes in the judicial field brought about by
AI
The Blue Book: As currently the world’s leading AI technology company, what
applications has iFlytek had in the judicial field? Are there any successful examples?
Liu Qingfeng: At present, iFlytek has in-depth exploration and extensive applica-
tions of AI in maintaining public security, promoting judicial collaboration, strength-
ening the deep application of big data, reforming the judicial system, etc. We have set
up artificial intelligence research and application labs for various political and legal
fields, with the Ministry of Public Security, the Supreme People’s Procuratorate, and
the Supreme People’s Court specializing in the construction of political and legal
informatization applications.
Today, iFlytek’s AI system has been applied in 25 provincial procuratorates and
31 provincial courts across China. The “AI-Assisted Case Handling System for Crim-
inal Cases” (referred to as the 206 System) jointly developed with the Shanghai High
Court has completed the formulation of an evidential standard for charges of 102
common criminal cases. The system can proofread, suggest, check and supervise
evidence in the handling of criminal cases by departments of public security, pros-
ecution, law enforcement and judicature. It has improved timely, accurate, overall
and scientific examination and judgment of evidence, avoiding “wrong start point,
wrong process, and wrong end”, has effectively prevented the occurrence of unjust,
false and misjudged cases, and has ensured judicial fairness. Presently, the system
has been efficiently adopted in Anhui, Shanxi, Guizhou, Xinjiang, Shenzhen and
other places.
The Blue Book: Through years of practice, what aspects of changes and
enhancements do you think AI technology can bring to the judicial field?
Liu Qingfeng: I am aware that AI changes the judicial field mainly in two aspects:
improving quality and efficiency. In terms of improving quality, judicial activities
have their own laws and characteristics, for instance, judicial impartiality, inde-
pendence, openness, and personal experience, as well as different experiences and
rational judgments of judges, prosecutors, investigators, etc., which have defined
them as the subjects of case handling. Meanwhile, AI is still in its infancy with
uncertainties and limitations. Therefore, its positions can be AI judge assistant, AI
prosecutor assistant, and AI investigator assistant. AI can only assist in case handling
and help to improve the quality but cannot replace judges, prosecutors, and inves-
tigators in dealing with cases. For example, it can standardize the procedures and
7 Expert Interviews on the Development of Artificial Intelligence … 283

standards for handling cases through unified guidelines of evidential standards; it

makes machine-assisted sentencing and similar case posting possible by constructing
evidential models for each type of case.
The role of AI in increasing efficiency can also be well understood. AI frees case
handlers from massive repetitive labor. Specifically, with OCR and intelligent cata-
loging, AI can automatically generate electronic case files in an effort to avoid manual
scanning files multiple times to complete them; AI can build a cross-departmental
collaborative case handling platform to deliver the whole process of case handling
online for the convenience of case handlers. The combination of AI with rule of
law in simple and repetitive legal affairs such as accessing files and referring to the
literature and laws, court records, and consultation demonstrates that substituting AI
for manpower can significantly enhance judicial efficiency and expand service scope
and thus help to modernize legal systems.
Difficulties related to data, technology, talent and systems are the main problems
that need to be solved in the judicial application of AI.
The Blue Book: During exploration, are there any difficulties in the application
of AI in the judicial field? What are the solutions to these difficulties?
Liu Qingfeng: We did encounter many difficulties. The first one is difficulties
of data. We acknowledge that data are essential for AI application. Law is a closed
normative system, and legal data need to be screened, which means that more samples
do not necessarily lead to better products. Like the literature, law is an embodiment of
excellence rather than quantity, so only after prior sorting and selection of legal data
by legal professionals and then with machine learning can fine products be developed.
At present, judicial organs’ data sharing ability is weak; however, barriers among
departments still exist, and there is a lack of relevant systems and legal support for
data sharing. A majority of historical cases in the judicial system are still recorded
in paper files, which costs much to digitize those files before they can be used as AI
training data.
The second difficulty is technology. We know there are three major stages of AI:
intelligent computing, intelligent perception, and intelligent cognition. At present,
intelligent cognition development is still in the initial phase, and AI has a limited
impact on complex judicial systems. It cannot solve all problems because the tech-
nology itself generates problems that need to be defined by legal professionals. For
example, a machine can easily identify case elements such as time, place and tools
of a crime, while currently, it fails to analyze specific circumstances, intent, and
reasoning of the case. The present approach is to make AI an assistive tool and
employ a “human–machine coupling” strategy to address comprehensive problems.
The next difficulty is the lack of interdisciplinary talent. Having data and tech-
nology alone is not enough to solve the problems in the judicial field. To develop
better products to empower the judiciary, interdisciplinary talent on AI technology
and legal knowledge are indispensable.
The last difficulty is institutional reform. The “206 System” mentioned earlier was
born under the requirement of “promoting trial-centered reform software for litigation
system”. However, when the system went online and launched trial operation, it
faced practical dilemmas, such as changing the habits of grassroots case handlers
284 Y. Cui

and strictly standardized case handling procedures, which resulted in their reluctance
to use the system. Therefore, on the one hand, we continuously rubbed shoulders
with the front-line case-handling personnel during the trial operation of the system
to optimize and transform it; on the other hand, we also cooperated with judicial
authorities to formulate a series of rules, regulations and assessment methods to
prompt case handlers to standardize procedures and improve the quality of case
handling, as well as promote fairness and justice. Hence, if AI is to be better applied
in the judicial field, in addition to optimizing products, improving relevant systems
and laws is also a crucial part.
From theory to practice: the training mode of interdisciplinary talents of AI
and law
The Blue Book: You have mentioned that society is in urgent need of interdisciplinary
talents of AI technology and legal knowledge. As a part-time professor in colleges
and universities and a person in charge of enterprises, how do you think we should
cultivate such talents?
Liu Qingfeng: In a word: from theory to practice. Colleges and universities need
to set up AI and law majors, concentrating on the integration of AI and traditional
disciplines, and nurture students with a solid theoretical foundation and innova-
tive ability. Enterprises should focus on industrial knowledge empowerment by AI
experts, closely follow the actual needs of the judicial industry in the process of
product development, and strengthen the industrial knowledge of professional talent
by hiring external business experts for empowerment training. Efforts should be made
to promote the establishment of joint laboratories between enterprises and judicial
authorities so that technology can help to enhance business.
In 2017, the Ministry of Education in China organized unicorn enterprises such as
iFlytek, Microsoft Research Lab—Asia, and Apple to jointly implement a number of
projects in the field of AI with universities, such as “High-quality Course Construc-
tion of Artificial Intelligence” and “Innovation and Entrepreneurship Education
Reform in Artificial Intelligence”. In 2019, Shanghai University of Political Science
and Law established the School of Artificial Intelligence and Law, of which I am an
honorary dean. Currently, we are working with the university to set up AI and law
courses and compile teaching materials.
The Blue Book: With deeper application of AI in the judicial field, what other
issues do you think we should pay attention to in the future?
Liu Qingfeng: Since the advent of the AI era, on the one hand, in the judicial
field, we should seize the opportunity to keep pace with the time, take the initiative,
accurately integrate judicial laws with AI features, and actively extend judicial appli-
cation so that AI can better serve the judiciary and promote judicial modernization
itself. On the other hand, the judiciary should address emerging issues raised by AI
in aspects such as law, security, employment, morality and ethics and governance,
strengthen forward-looking research on AI development and rule of law, proactively
build future AI and rule of law system, guarantee the sound and sustainable devel-
opment of AI with rule of law, and provide legal protection to implement national
AI strategy.
7 Expert Interviews on the Development of Artificial Intelligence … 285

Values pursued by traditional laws remain valid, but AI affects how we realize
legal values. Therefore, with widespread application of AI, it is more important
to keep a clear head, reflecting on the legal, ethical and social issues involved, to
propose conservative and balanced institutional strategies and avoid technological
determinism.

5 Rule of Law Guarantee for Deeper Application of AI

in Healthcare—Interview with Fan Xianqun, Deputy
Secretary of the Party Committee of Shanghai Jiao Tong
University, Secretary of the Party Committee of Shanghai
Jiao Tong University School of Medicine, and Dean
of China Hospital Development Institute of Shanghai
Jiao Tong University

Fan Xianqun: As a professor and discipline leader of the Ophthalmology Department

of Shanghai Ninth People’s Hospital, Shanghai Jiao Tong University, Fan has been
devoting his efforts to clinical treatment and basic research on orbital diseases and
ocular tumors. He was awarded the Fellowship of Royal College of Ophthalmologists
(UK) in 2015 and a Fellowship Ad Hominem by the Royal College of Surgeons of
Edinburgh in 2019 as President of Asian-Pacific Society of Ocular Oncology and
Pathology. Fan was named a Distinguished Professor of the “Chang Jiang Scholars
Program” by the Chinese Ministry of Education, a member of the Thousand Talents
Program, Young and Middle-aged Experts with Outstanding Contributions by the
National Health Commission of China, and Shanghai Leading Talent. He is currently
Deputy Secretary of the Party Committee of Shanghai Jiao Tong University, Secretary
of the Party Committee of Shanghai Jiao Tong University School of Medicine, and
Dean of China Hospital Development Institute of Shanghai Jiao Tong University.
Healthcare is one of the fields in which AI is most deeply applied
The Blue Book: As an expert in clinical medicine, what is the current stage of AI
application in healthcare, according to your understanding?
Fan Xianqun: Currently, healthcare is one of the areas where AI is most deeply
applied. According to statistics in the Blue Book of World AI Industry Develop-
ment (2018) coreleased by the China Academy of Information and Communications
Technology and Gartner, the healthcare sector ranks first among AI vertical appli-
cations in China, at as much as 22%. AI empowers healthcare, assisting clinicians
in diagnosis or treatment, for example, by improving the efficiency and treatment
of medical institutions and personnel, optimizing the treatment process, facilitating
patient access, significantly reducing medical costs, and enabling people to achieve
scientific and effective daily health check-ups and better manage their own health.
The earliest applied AI in healthcare was the AAPHelp system developed by the
University of Leeds in 1972 to assist in the diagnosis of sharp abdominal pain and
286 Y. Cui

offer surgical assistance. Starting in the twenty-first century, since AI as a whole

entered the industrial development stage; AI + healthcare applications ushered in a
period of rapid development because it is easy to land in the healthcare sector. The
Watson robot, an AI system developed by the DeepQA Research Team from IBM
in 2007, can quickly give diagnostic tips and treatment advice by asking patients
about their medical history and symptoms and using AI to process and analyze
natural language. In 2017, the U.S. Food and Drug Administration (FDA) approved
CardioDL, an AI software used to analyze cardiac magnetic resonance imaging, and
adopted the world’s first AI medical facility IDx-DR in 2018. China’s development
of AI applications in healthcare has been keeping pace with the world. For instance,
machine vision in ophthalmology for medical image recognition and lesion iden-
tification and AI fundus screening have been accepted by hospitals and grassroots
organizations over more than a decade of technological updates and industrial land-
ings. Since 2016, Peking Union Medical College Hospital has started explorations
and research on fundi related to AI and achieved results in diabetic retinopathy.
Professor Liu Yizhi of Zhongshan Ophthalmic Center, Sun Yat-sen University led
cooperation between the university and a research team from Xi’an University of
Electronic Science and Technology, based on deep learning algorithms, establishing
the “Congenital Cataract-Cruiser (CC-Cruiser) Artificial Intelligence Platform” and
“AI Application Clinic”, where clinicians are assisted by AI cloud platform in diag-
nosis and treatment. The ophthalmology team of Shanghai Ninth People’s Hospital
of Shanghai Jiao Tong University created a method to automatically diagnose orbital
fracture based on a deep convolutional neural network by accurately labeling orbital
fracture image data. Professor Ye Juan’s team from the Ophthalmology Department
of the Second Affiliated Hospital of Zhejiang University School of Medicine devel-
oped an approach to automatically identify and diagnose malignant eyelid tumors in
pathological images based on computer deep learning.
The Blue Book: The Endo-navigation system for orbital surgery developed by
your team won the second prize of the National Science and Technology Progress
Award in 2015. As a well-known medical school in China, are there similar AI
medical applications in Shanghai Jiao Tong University School of Medicine?
Fan Xianqun: Shanghai Jiao Tong University School of Medicine attaches great
importance to the development of AI in healthcare and carries out exploration
and research on corresponding technologies, policies, ethics and laws. In indus-
trial studies from 2018, the Artificial Intelligence Institute of Shanghai Jiao Tong
University, Shanghai Health Development Research Center and Shanghai Jiao Tong
University School of Medicine jointly published the White Paper on Artificial Intel-
ligence in Medicine annually to comprehensively analyze the current situation, chal-
lenges and trends of AI in the healthcare industry in major countries around the world
and provide intellectual support for China’s AI healthcare industry.
In technology exploration, the Medical School of Shanghai Jiao Tong Univer-
sity has actively developed AI medical technology, medical image big data, image
recognition and other assistive diagnosis and treatment systems, applied them in all
affiliated hospitals of the medical school, and achieved good results. Specifically,
the Institute of Medical Robotics of Shanghai Jiao Tong University together with the
7 Expert Interviews on the Development of Artificial Intelligence … 287

affiliated Ninth People’s Hospital has formed four R&D teams for orbital surgery
robots, craniomaxillofacial osteotomy robots, jaw reconstruction robots and auditory
implant surgery robots.
I myself have long been dedicated to the diagnosis and treatment of orbital diseases
and ocular tumors, and I have led a research team to develop an endoscopic navigation
system for orbital surgery, which has improved the safety and effectiveness of orbital
surgery, but it is not true AI. Based on this, we are further developing an augmented
reality-integrated robotic system for orbital surgery under monitoring of endoscopic
navigation, which is expected to perform more precise orbital surgery.
The team of Professor Wu Hao from the Department of Otolaryngology-Head and
Neck Surgery of Shanghai Ninth People’s Hospital, in collaboration with Shanghai
Jiao Tong University and enterprises, has carried out independent research and devel-
opment of domestic auditory implant surgery robots and has initially completed the
design and prototype production of otologic surgery robots, artificial cochlear elec-
trode implantation devices and precurved electrode implantation devices, which are
expected to achieve clinical transformation for the benefit of patients with severe
hearing impairment. Children’s Hospital has deployed and tuned two robots since
2018 to automate the entire process of delivering intravenous infusion kits and drugs
to pharmacy intravenous admixture services and inpatient pharmacies. In 2019, the
research group of Professor Yang Junlin from Xinhua Hospital Affiliated to Shanghai
Jiao Tong University School of Medicine announced an AI screening system for scol-
iosis based on the appearance of human bare back, which is accurate as average human
experts. The system is expected to be used for large-scale scoliosis screening and
makes it possible to track the trajectory of scoliosis changes in patients. Meanwhile,
during 2020 COVID-19, Professor Wu Tao’s research group at Shanghai Jiao Tong
University School of Medicine donated AI medical service robots to three hospitals
in Wuhan, providing material support for AI in healthcare in epidemic-stricken areas.
Prospects and main legal issues for the application of AI in healthcare
The Blue Book: As far as you know, how does the healthcare industry view the
application of AI and challenges it brings to the rule of law?
Fan Xianqun: The healthcare industry is looking forward to AI development.
Since AI enters an era of machine learning, the future will be a period of rapid devel-
opment for AI to further integrate with key technologies and industries in healthcare.
The application of AI in healthcare enhances medical service ability and efficiency.
On the one hand, it can facilitate patient access and significantly reduce the burden
on doctors to share high-quality medical resources; on the other hand, it can also
boost medical and health services and improve the efficiency and effects of diag-
nosis and treatment. On the whole, the application of AI in healthcare is maturing,
and there will be extensive application and development in aspects such as medical
research, intelligent diagnosis and treatment, and family health management in the
future. Shanghai has long been the first-tier city in terms of the talent population in AI
healthcare and policy and financial support for research capabilities. When actively
promoting graded diagnosis and treatment, Shanghai advocates the establishment and
improvement of a family doctor system, in which hospitals render contracted services
288 Y. Cui

of family doctors via the Internet so that AI in healthcare will experience sustainable
growth in the provision of primary medical services. The government should start
to increase investment in AI in healthcare, value talent training, encourage hospitals
to strengthen the application of AI in healthcare, and emphasize the combination of
industry, academia and research so that AI in healthcare can better serve the people
and contribute to the overall healthcare in Shanghai and in China as a whole.
However, the application and practice of AI in healthcare brings new ethical
and legal challenges, such as attribution and privacy protection of AI healthcare
data, technology abuse and safety, ethical regulation, and subject to AI healthcare
liability. Therefore, while encouraging the development of AI medical technology
and industry, there is an urgent need to enact laws on AI in healthcare, strengthen
supervision of the industry, clarify subject to liability, develop ethical and technical
principles to regulate medical practices, protect big data and personal privacy, and
guarantee medical quality and safety. Sound laws have been promulgated to protect
patients’ rights to health and privacy under the current legal framework, and provi-
sions of such rights have been prescribed in the Civil Code of the People’s Republic of
China, Law of the People’s Republic of China on Basic Medical and Health Care and
the Promotion of Health, Law of the People’s Republic of China on Medical Practi-
tioners, etc. In the meantime, regulations concerned with AI in healthcare have been
approved. As an illustration, China issued the Management Regulation for Artificial
Intelligence Assisted Therapy Technology as early as 2009, with an amendment in
2017 that stipulates the minimum requirements for medical institutions and personnel
to perform AI-assisted therapy, regulates the medical practice of AI-assisted therapy,
and ensures medical quality and safety.
The Blue Book: What is the most critical issue of rule of law in the development
of AI in healthcare? What are problems that need to be solved?
Fan Xianqun: The primary issue in the regulation of AI in healthcare is to deter-
mine subject to liability of AI healthcare performance. It should be a doctor rather than
algorithms that make the final decision, so AI medical practice should be regarded
as the doctor’s treatment. In the long run, AI will have a profound impact on medical
science and will change the medical practice, patient experience, routine proce-
dures of doctors and how medical personnel perform their duties. Nevertheless, AI
is ultimately created by humans, and its role is to better serve humans.
On the one hand, the widespread application of AI in healthcare will produce
a technical “black box” in which doctors need to fully understand the diagnostic
process and result and be responsible for dealing with concerns of patients, which
is more than just pressing a button. On the other hand, although AI in healthcare
cannot replace doctors, it can take their repetitive and regular labor, thus improving
efficiency. As prescribed in the Regulations for the Supervision and Administration
of Medical Devices, China’s current AI medical equipment, instruments, computer
software, etc., are managed as medical devices, including the da Vinci robotic surgical
system and ET medical brain AI system. Adopting such devices to diagnose and treat
diseases is generally considered to be professional medical means and measures in the
scope of medical technology. Therefore, it should be clear that medical institutions
7 Expert Interviews on the Development of Artificial Intelligence … 289

and doctors that use or will use AI devices in AI medical practices are the subjects
of medical conduct and legal liability.
Legislation: the most effective mechanism to facilitate the development of AI
application in healthcare
The Blue Book: It is well known that the medical industry has a huge amount
of medical data that involves personal privacy and national interests. In terms of
medical information protection, what issues are confronted in the application of AI
in healthcare?
Fan Xianqun: Problems of AI healthcare data mainly entail big data ownership
and security. Sharing medical big data pursues public interest, and its potential value
must be realized with a prerequisite for fully protecting individuals’ rights. However,
with the introduction of AI technologies such as data collection, data storage, data
application, and machine learning, disputes have arisen over the ownership of medical
data, namely, intellectual property rights in medical databases. In practice, generally,
medical institutions and data companies clearly define the ownership of property
rights and the resulting benefit sharing in the contract. Therefore, medical institutions
should be aware of property rights and highlight the protection of intellectual property
rights of big data when developing data with third-party companies.
Data security is a key issue in the application of medical big data. Since there has
been leakage of statistics and human genome data caused by AI training with patient
data in China, the country’s present laws and regulations emphasize the protection
of data and personal privacy. Article 1226 of the Civil Code of the People’s Republic
of China stipulates that medical institutions and their medical staff should keep their
patients’ private information and personal information confidential. Anyone who
divulges the private information or personal information of a patient or discloses his
medical records without the patient’s consent shall bear tort liability. Article 49 of
the Law of the People’s Republic of China on Basic Medical and Health Care and
the Promotion of Health provides that the state promotes health informatization for
the whole people, advances the application and development of health and medical
big data and artificial intelligence, accelerates medical and healthcare information
infrastructure construction, develops technical standards on the collection, storage,
analysis and application of medical and health data, and promotes the populariza-
tion and sharing of high-quality medical and healthcare resources through informa-
tion technology. Such measures have promoted the establishment and improvement
of systems of medical information exchange and security by medical institutions,
applied information technology to provide remote healthcare, and created an inte-
grated online and offline model of healthcare services. At the same time, Article 23 of
the Measures for the Administration of Internet Hospitals (for Trial Implementation)
has corresponding provisions of medical data as well that Internet hospitals shall
strictly implement laws and regulations relevant to information security and confi-
dentiality of medical data, properly store patient information, and shall not illegally
trade or divulge patient information. In the event of disclosure of patient informa-
tion and medical data, medical institutions should promptly report to administrative
departments of health in charge and immediately take effective countermeasures.
290 Y. Cui

Although the law clearly prohibits disclosure of patient privacy and contains protec-
tive provisions of data security, most of these provisions regulate medical institutions,
with a lack of basic institutional norms and standards for the use and development
of data in medical big data companies, which requires refinement of management
details.
The Blue Book: What role do you think medical ethics should play in AI
application in healthcare?
Fan Xianqun: At present, the National Medical Products Administration has
issued guidance and standards for AI in healthcare and has initiated certification
processes in terms of scope, risk, and clinical trials. In October 2019, the National
Medical Products Administration released an announcement that approved the estab-
lishment of a department in charge of the standardization of AI medical technology
devices, and the Standardization Research Group was created specifically for the
standardization of AI medical devices. However, regulations on the corresponding
ethical review mechanism for AI in healthcare are lacking. According to Article 32 of
the Law of the People’s Republic of China on Basic Medical and Health Care and the
Promotion of Health, clinical trials on drugs or medical devices and other medical
studies shall comply with medical ethics, pass ethics review and obtain informed
consent in accordance with law. In reality, some hospitals introduce ethical review
mechanisms for drug clinical trials to conduct ethical review of AI in healthcare and
sign patient informed consent forms when applying medical AI systems to patients to
protect their right to informed consent and treatment choice, as prescribed in Article
32.
I believe that ethical regulation on AI in healthcare should be people-oriented,
strengthen medical quality and patient safety, and protect patient privacy and that the
application of AI technology should absolutely respect patients’ right to informed
consent and prevent abuse, with an ultimate goal of serving medical diagnosis, treat-
ment, rehabilitation, and health management and guarantee patients’ right to health
and dignity.
Efforts should be made to develop medical science and AI in coordination and
to improve the legal imputability system
The Blue Book: There have been eight effective cases of medical disputes related
to da Vinci robots in China that fall into the category of medical malpractice claims.
Laws applied by judges mainly focus on damages in medical product liability
provided in the Tort Law of the People’s Republic of China. Technologies such
as AI-assisted film reading and AI-assisted diagnosis sprint forward, but the law
lags behind. A pathology report directly involves removal of organs or limbs of
patients, and mostly it results in material medical incidents once there is an accident.
Reducing the number of pathology medical-legal disputes and responding to such
disputes properly are issues that need to be addressed. How do you view AI medical
disputes in light of the development of AI in healthcare?
Fan Xianqun: I have mentioned two points earlier. First, AI medical practices
should be recognized as the medical performance of medical institutions and their
personnel. Second, AI medical equipment and technologies should be managed as
7 Expert Interviews on the Development of Artificial Intelligence … 291

medical devices and technologies. Article 1223 of the Civil Code of the People’s
Republic of China stipulates that where damage is caused to a patient due to a defect
in a drug, disinfection product, or medical instrument, the patient may claim compen-
sation against the drug marketing license holder, the manufacturer of the drug, or
against the medical institution. This provision provides that the principle of no-fault
liability applies to damage medical products such as medical instruments. That is,
where medical products cause damage to a patient, the patient may claim compen-
sation against the medical institution concerned, regardless of medical malpractice
by such institution or personnel. In the case of medical damage and torts caused by
special AI medical products such as the da Vinci surgical robot, no-fault liability for
medical damage is currently applicable. However, in judicial cases, the ascertain-
ment of causation and fault in the medical field largely relies on the application of
identification. Employing AI technology further requires experts with both medical
and AI background knowledge to apply their expertise, skills and experience to assist
judges in making judgments on factual issues to properly handle cases and ensure
fairness of the adjudication. Most of the current appraisal conclusions focus on iden-
tifying malpractice in medical institutions, with a shortage of detection and analysis
of medical product defects. Such appraisal conclusions often make medical institu-
tions assume compensatory liability to patients, although tort liability also provides
that medical institutions are entitled to recover from the responsible producer after
compensation. According to Article 22 of the Judicial Interpretation on Medical
Malpractice Damages, the product producer is jointly and severally liable with the
medical institution, but in practice, it is difficult to recover from the enterprise that
produces AI devices.
AI medical imaging robots are devised as medical devices and software systems
that assist doctors in disease diagnosis and treatment. Medical imaging doctors
perform medical imaging and radiological examinations on patients upon appli-
cation by clinicians and output preliminary diagnostic reports through acquisition,
recognition, and structured processing of images. For example, the Watson system
developed by IBM in 2016, a well-known company in global AI medical diagnosis
technology, provides oncology solutions that have served multiple medical institu-
tions in China, with a diagnostic accuracy of up to 90%, which is much higher than
that of lung cancer diagnosis by human doctors. However, there is still a 10% misdi-
agnosis. Who shall be liable if the 10% misdiagnosis leads to medical disputes? At the
present stage, the final results of AI in healthcare still need manual verification and
review. Based on the preliminary diagnosis report of AI medical images, radiologists
apply their expertise and experience to write a final diagnosis report for clinicians’
reference in confirming diagnosis and treatment. Thus, doctors are responsible for
the patient’s treatment outcome. In determining whether there is diagnostic fault, the
principle of determining the average level of diagnosis and treatment shall prevail,
which is stipulated in Paragraph 3, Article 1224 of the Civil Code of the People’s
Republic of China that a medical institution shall not assume liability for compen-
sation for any damage caused to a patient during diagnosis and treatment when the
diagnosis and treatment of such a patient is beyond the average medical level at that
292 Y. Cui

time. Stipulation and application of such disclaimers, in my opinion, is conducive to

the innovation and development of AI medical technology.
The social value brought by AI technologies in the medical field, such as da Vinci
surgical robots, AI-assisted film reading and AI-assisted diagnosis, is incredible with
broad application prospects. The liability of AI in healthcare damage overemphasizes
that the protection of personal interests will affect the progress and development of
AI medical imaging technology and that the protection of public interests will make
the underprivileged more vulnerable and intensify social conflicts, so we need to
think comprehensively. When AI application causes medical disputes and ethical or
legal conflicts about people, a reasonable system of liability and compensation at
the AI application level needs to be created to ensure that AI in healthcare plays its
proper clinical value.
Cultivating interdisciplinary medical talent and law by strengthening the legal
literacy of AI medical talent
The Blue Book: As a prestigious university in China, does the School of Medicine
of Shanghai Jiao Tong University have a vision for AI medical talent training?
Fan Xianqun: In April 2018, the Ministry of Education studied and formulated
the Action Plan for Leading AI Innovation in Higher Education and researched the
establishment of AI majors to further improve the AI discipline system in univer-
sities. According to the setting of the undergraduate major catalog of the Ministry
of Education, majors related to AI and healthcare at Shanghai Jiao Tong Univer-
sity primarily include biomedical engineering, medical informatics engineering, and
bioinformatics, a subdiscipline of biology. An undergraduate-master-doctoral talent
training system was initially formed. However, from the perspective of professional
settings, the knowledge structure is relatively single and fails to meet the require-
ments of knowledge and ability structures related to AI and healthcare. To better
promote the cultivation of AI healthcare talent with respect to professional construc-
tion, we should actively adapt to the new trend of global scientific and technolog-
ical revolution and life sciences development and establish new majors such as AI
medical engineering, medical artificial intelligence, and Internet + healthcare to
accelerate the cultivation of innovative medical talent with multidisciplinary knowl-
edge. In terms of curriculum, students will learn interdisciplinary courses in law,
management, humanities and medicine, such as Health Law, Social Medicine and
Health Management, Medical Sociology and Health Economics, to enhance their
medical and legal literacy. With regard to teaching methods, we will strengthen
student orientation and independent learning and insist on the combination of theo-
retical and experimental teaching, classroom teaching and extracurricular indepen-
dent learning. We will adopt case-based learning (CBL), problem-based learning
(PBL), team-based learning (TBL) and competency-based learning (CBE) according
to the characteristics of each curriculum phase, focus on enhancing students’ under-
standing and mastery of basic medical theoretical knowledge, improve their analyt-
ical thinking, teamwork and job competence; concerning teaching evaluation, we will
refine the formative-summative evaluation system, and build a diversified education
supervision and evaluation system to ensure scientific, objective and fair teaching
7 Expert Interviews on the Development of Artificial Intelligence … 293

quality assessment, encourage advancement of teaching quality, and speed up the

construction of “New Medical Science”.

6 Grant Legal Personality to Artificial Intelligence

and Pursue a Quest for AI Data Governance—Interview
with Klaus Heine, Director and Professor of Erasmus
Graduate School of Law at Erasmus School of Law,
Erasmus University Rotterdam

Klaus Heine: Professor of Law and Economics and the director of Erasmus Graduate
School of Law at Erasmus School of Law, Erasmus University Rotterdam. He has a
Jean Monnet Chair of Economic Analysis of European Law and is a director of the
Jean Monnet Centre of Excellence on Digital Governance as well as a director of the
Erasmus Centre for Data Analytics.
The Blue Book: Hi Professor Heine, very glad to have the interview with you.
I know you were invited to give a keynote speech at the Rule of Law Forum of the
2nd WAIC in Shanghai last year. How about your experience?
Klaus Heine: Thank you very much. It is also my great honor. It was wonderful! I
would like to thank the perfect organization committee of the WAIC, Shanghai Law
Society and all volunteers. This is an excellent conference and I was impressed.
The Blue Book: The title of your speech is “The Quest for AI Data Governance”.
One of the most important ideas is that you propose granting legal personality to
robots. Why would that be a good idea?
Klaus Heine: Differently from what many nonlawyers may think legal personality
is not something that is reserved for human beings. Additionally, firms have their own
legal personality, e.g., in corporate form as limited liability companies. In addition,
if we go back in time, for example into the middle ages or to ancient China, Rome or
Greece, also humans were not fitted with the same legal personality as we are used
to it today. Slaves had rights but not the right of self-ownership, or insolvent citizens
lost the right of self-determinacy until they had paid back their debt. What I will
say is that the concept of legal personality is a legal artifact, a set of manmade rules
that shall serve specific purposes. Typically, a legislator would like to shape legal
personality in a way that it is to the benefit of society: Legal persons shall have an
incentive to prevent harm, shall raise capital for innovations, shall distribute profit
between stakeholders, etc. Generically, legal persons are distinctive spots that have
been granted a specified autonomy to make decisions and that are liable for their
actions. Hence, legal persons may be the owner of property and may be a partner in
a contract with corresponding obligations.
The Blue Book: How does innovation relate to lawmaking?
Klaus Heine: The creation of new forms of legal personality is a reaction to the
challenges of new situational conditions and may be the necessity to raise capital for
large infrastructures (train projects in the eighteenth century), the exploration of new
294 Y. Cui

continents (European colonies as of the seventeenth century) or the advancement

of SMEs as of the twentieth century by private limited companies. Technological
drivers are triggering the legal system to come up with new materializations of legal
personality, legal personalities that can better deal with actual challenges than the
incumbent legal personalities which were meant to cope with the challenges from
the past. Granting legal personality to AI is therefore a natural step to integrate
autonomous decision making of machines into our society and to let AI do what
society likes AI to do for it: Creating wealth and health for all people.
The Blue Book: However, is a modest adaptation of the law that we actual have
not sufficient? A gradual modernization of the law would also be much easier to
achieve, and the legal profession is much more likely to accept it.
Klaus Heine: In my view, AI will prosper indifferently from the legal regime
around it. AI is a very strong technological imperative that does not care overly
what legislators decide or what the doctrinal legal standpoint of law professionals
is—at least not in the medium and the long run. Those jurisdictions or countries
that master adapting their legal infrastructure complementary to the technology will
have an enormous competitive advantage and can reap the benefits of AI earlier than
late adopters. It is the coevolution of technology and institutions that is ultimately
decisive for economic growth.
The Blue Book: So in your opinion, what does the AI governance framework
look like?
Klaus Heine: The paradigmatic shift lies in the direct attribution of rights and
obligations to AI. At the moment an AI is treated as any other machinery belonging
to a natural person or company. Rights and obligations concerning an AI are only
addressed to the legal person who owns it. This means that the full property rights of
AI rests unspecified with the owner. AI is—so to say—covered behind the corporate
veil of its owner. It is the owners’ interests that play out against the public interest
when it is about the regulation of AI. This two-dimensional relation only changes
when AI obtains its own legal personality. Then, it still belongs to the company but
can directly be accessed by the legislator or regulator. This means that AI obtains a
specific contractual status within the company as employees, management or share-
holders have. The specific status of AI allows regulators to directly govern the relation
between AI and society.
The Blue Book: Could you give an example about the way in which granting legal
personality to AI might be advantageous for society, and how it may solve pending
legal problems of AI?
Klaus Heine: For example, large pharmaceutical companies may employ AIs to
create new drugs. At the moment, the new receipts of the drugs accrue automatically
as patents to the company. These companies may then use intellectual property rights
to hinder new companies from entering the market. Incumbent competition law is not
well prepared to deal with those or similar cases because it has to regard hard wired
intellectual property rights law as it is. However, if intellectual property rights are
associated with AI, it would be possible to expropriate AI rather than the company
from the Intellectual Property Right. Historically, intellectual property rights were
meant to incentivize and compensate human entrepreneurs for risk taking. That is not
7 Expert Interviews on the Development of Artificial Intelligence … 295

the case with AIs. Thus, it is sufficient to guarantee companies and investors make
a profit from employing an AI but not giving them the right to deny access to the
data and intellectual achievements generated by AI. AIs fitted with legal personality
may have to pay taxes, may have to insure against accidents, etc. A whole new world
of legal possibilities opens when one makes the paradigmatic step to grant legal
personality to AIs and starts thinking about the design of the contractual nexus with
which AIs become accommodated into our society. I do not see yet how society could
achieve the same beneficial results without changing the legal status of AI. Those
jurisdictions will have a competitive advantage that first moves to a new legal status
of AI.
The Blue Book: Thank you, professor, I know in the past several months you
have been quite focusing on this topic. What have you achieved after you came back
from Shanghai? What is your plan for this year?
Klaus Heine: My pleasure. After I came back from China, as the director of
DIGOV,6 I devoted myself to enhancing the conversation among scholars from
different countries. On January 28 to 30, 2020, I organized a kick-off conference
on digital governance. Scholars across European countries got together discussing
topic issues relating to AI and data science. In addition, I am now also leading
the Erasmus Centre for Data Analytics. In this way, we conduct multidisciplinary
research with data scientists, economics and the people industry so that we will have
a closer look at how data and AI are changing our society. In addition, I promoted
joint research with Chinese scholars. We hope to reach some agreements on digital
governance between China and Europe.

7 The Civil Code Should Prepare for New Types of Tort

Brought About by Artificial Intelligence—Interview
with Zhang Xinbao, Editor-in-Chief of China Legal
Science

Zhang Xinbao: Distinguished Professor of the “Chang Jiang Scholars Program” at

China Law School of Renmin University, Editor-in-Chief of China Legal Science,
President of the Research Institute of Legal Journals, Distinguished Professor of the
Changjiang Scholar Award Program of the Ministry of Education, and one of the
National Outstanding Young and Middle-Aged Jurists.
The Blue Book: Hello Professor Zhang, as the Chief Editor of China Legal
Science, you have long been concerned about the dynamics of legal research in
China. In your opinion, what are the main issues of AI legislation that scholars have
focused on since 2019? What are new hotspots identified and trends emerged?
Zhang Xinbao: According to a preliminary estimate by scholars, in 2019, nearly
21 major law journals published 111 legal papers covering topics such as civil legal

6 DIGOV is the Jean Monnet Center on Digital Governance financed by the European Union.
296 Y. Cui

subject qualification of AI entity, attribution of intellectual property of AI-generated

objects, criminal liability of AI, protection of personal information and privacy,
algorithmic justice, intelligent justice, AI contracts, legal issues of application of
blockchain technology, etc. I believe these topics will remain hot in 2020’s research.
With the promulgation of the Civil Code of the People’s Republic of China, IT-
relevant provisions therein, especially the protection of personal information, tort
liability, the protection of online virtual assets and data property, online transactions,
and obligations and liabilities of the platform, will attract more attention from the
legal community.
The Blue Book: The compilation of the Civil Code is a major event in the modern-
ization of China’s national governance system and capacity. How do you view the
increase of tort liability for personal information infringement and the construction
of necessary class action system in the Civil Code?
Zhang Xinbao: The Civil Code provides for the protection of personal infor-
mation in principle in Article 111 and the protection of privacy and personal infor-
mation in the Book of Personality Rights. The Civil Code stipulates the liability for
infringement upon civil-law rights through the network in the Book of Tort Liability,
especially detailed rules for network service providers where bearing tort liability. In
cases of infringement of personal information, there is a huge gap in the ability of the
burden of proof prove between the perpetrator and victim. The perpetrator is often a
large enterprise with professional knowledge and technology, while the victim is a
natural person with little or without such resources. This serious imbalance in ability
to prove is very similar to the situation of ecological torts. Therefore, reversing the
burden of proof in the domain of personal information would profoundly strengthen
the protection of personal information. In addition, it is difficult to quantify damage
caused by infringement of personal information, and the scope (amount) of damage
that an individual victim can prove is completely disproportionate to benefit that
the perpetrator may obtain as a result of the infringement. The traditional system of
damage is hardly an effective deterrent against infringements of personal informa-
tion. It is believed that if a punitive damage system could be applied in this area,
deterrence and prevention would be greatly enhanced.
“Class Action” is a type of civil lawsuit governed by procedural laws. In accor-
dance with the current Civil Procedure Law of the People’s Republic of China, an
action may be brought by a representative elected by the infringed, which is similar
to the “representative action” provided for in some countries.
The Blue Book: At present, there are numerous cases of alleged infringement of
online data and theft of online virtual assets. How should we increase the liability
for such infringement after the issuance of the Civil Code to better protect interests
concerned?
Zhang Xinbao: Article 127 of the Civil Code prescribes the protection of data and
online virtual assets, without more specific provisions of protecting data (including
big data) and online virtual assets in the subparts. However, since online virtual
assets and data are recognized as property protected by civil law, provisions of the
Book of Tort Liability shall prevail in the event of infringement, and the infringer is
undoubtedly liable for civil liability such as making restitution.
7 Expert Interviews on the Development of Artificial Intelligence … 297

Moreover, the Intellectual Property Law and the Anti-Unfair Competition Law of
the People’s Republic of China also provide protections of data and online virtual
assets, and such provisions can be invoked in some cases to safeguard these rights and
interests. Now, in the surge of the Internet economy, legal issues arising from various
new forms of business are bound to be endless. In the face of these problems, on the
one hand, we need to introduce new legislation to provide incremental institutional
supplies. However, on the other hand, we should consider whether current laws can
solve such problems as well, that is, settle new disputes with the existing legal system.
The Blue Book: How to integrate date and the collection and processing of
personal information into AI? Why you suggest that the use of big data should
be “free + charged”?
Zhang Xinbao: AI’s basic principle is to collect and process a large amount
of personal information to extract people’s way of thinking, behavior and decision
making so that the AI system can learn a lot and in depth to think, behave and make
decisions like people, and make products or render services based on this. Therefore,
the collection of a myriad of personal information (data) is the key to AI success and
high-level intelligence. From this point of view, data and the protection of personal
information (data) are highly valued by the information industry and are regarded as
“industrial raw material” of the information age.
It is also because data are so important that innumerable operators currently use
apps to excessively collect personal information or use it for purposes to which
they have not been given user consent. We advocate that individuals strengthen the
protection of personal information, sensitive personal information in particular, to
strike a balance of maximum use of data after depersonalization (deidentification).
Thus, universal free access to web services provided by apps and exceptionally paid
services (navigation services, for example) is one way to keep balance.
The Blue Book: Currently, services of self-driving taxis have been launched in
Shanghai, Guangzhou and other places. The liability rules regarding tort brought
about by AI technology applications such as autonomous driving are a hot topic of
social concern. How do you think the security obligations and tort liability of network
platforms such as ride-hailing can be determined?
Zhang Xinbao: At present, there are different theoretical views on the tort liability
brought about by AI technology products such as autonomous driving, and there are
no specific provisions concerned in the Book of Tort Liability of the Civil Code.
From the existing statutes, product liability can be applied in some cases. The issues
involved are complex, the technology is still being worked out, and legal norms
concerned need to be established when the technology is ripe.
In the Book of Tort Liability in the Civil Code, there are special provisions on the
liability of nonoperating “goodwill” hitchhiking, but there are no provisions on the
liability of operating ride-hailing, which causes injury. I believe that for the latter, if
the platform itself is also an operator of the ride-hailing, certainly it should bear tort
liability of the operator; if it is not the direct operator, but charges a corresponding
percentage of the fee, it should bear the tort liability where it fails to fulfill the duty
of maintaining security; if it is an information provider rather than an organizer, and
298 Y. Cui

does not contribute to information exchange for the operator, then it is not at fault
and thus does not bear corresponding liability.

8 Promoting Rule of Law and Modernization of Cyber

Governance System and Governance
Capacity—Interview with Du Qian, President
of Hangzhou Internet Court

Du Qian: President of Binjiang District People’s Court in Hangzhou, President,

member of the Judicial Committee; Judge of Hangzhou Railway Transport Court;
President, member of the Judicial Committee, and Judge of Hangzhou Internet Court
since August 2017.
The Internet Court provides judicial guarantees to promote the deep integration
of the Internet, economy and society
The Blue Book: Hello, President Du. Both the judicial field and social media are
very interested in the Hangzhou Internet Court where you are working. Can you
briefly introduce the achievements made by the Hangzhou Internet Court since its
establishment?
Du Qian: The Hangzhou Internet Court has been closely oriented to the require-
ments of reform of central government, actively promoting judicial reform and inno-
vation, and enhancing the rule of law in cyber governance since its establishment.
As of May 7, 2020, a total of 94,305 disputes of all kinds were filed to the Hangzhou
Internet Court, and 92,177 were resolved. The online trial mode is 100% applicable
to plaintiff-related cases. In 2019, judges settled 1,181 cases per capita, the highest in
Zhejiang Province. Judicial credibility of the Internet has been generally recognized:
it was included in Memorabilia since the 18th CPC National Congress by Central
Government, named as one of the 40 “firsts” over 40 years of reform and opening
up. Judicial credibility of the Internet appeared twice at the Digital China Summit
and was selected as the Annual Best Practice Achievement of the first Summit. It has
been highly evaluated by leaders of national ministries and commissions, scholars,
experts and media reporters and has been hailed as a milestone in the judicial field.
The Hangzhou Internet Court concentrates on the strategic deployment of building
the country with a strong network and provides judicial guarantees to promote the
deep integration of the Internet and economy and society. By handling emerging,
difficult and complex cases, the Court has set up norms, guiding behavior and value
judgments by adjudication of individual cases, which plays a guiding role of judicial
adjudication to ensure innovation-driven development, maintain honest transactions
on the Internet, prevent and resolve Internet financial risks, create a clear cyberspace
and protect social public interests. For example, the first public interest litigation
case on the Internet that promotes core socialist values ended, in which the defendant
who defamed heroic martyrs Dong Cunrui and Huang Jiguang was ordered to stop
7 Expert Interviews on the Development of Artificial Intelligence … 299

sales and make public apology at national media. The case was given overwhelming
response from society, with Political and Legal Affairs Committee of CPC Central
Committee, Supreme People’s Court, Communist Youth League of CPC Central
Committee and People’s Daily reprinting the report, and over 10 million people
paid attention to this case on the day of pronouncing sentence. Hear the first case
of intellectual property that concerns blockchain technology and fixed electronic
evidence in China, create evidence collection process, identification standards and
operation rules of Internet evidence. The first case of unfair competition dispute over
big data products was closed, with a ruling that network operators have independent
property rights and interests of the data products developed by them. The case was
selected as one of the “Top Ten Intellectual Property Cases” and “Top Ten Civil and
Administrative Cases of the Year”. At the same time, it strengthened the research
and application of judicial big data, and publications such as the Internet Judicial
Development Index, White Paper on E-Commerce Cases, Big Data Analysis Report
on Internet Financial Trials and Report on Judicial Protection of Internet Copyright
were released to improve institutional supply and highlight the role and function of
the Internet Court in promoting the rule of law in online social governance.
“Traces are left and traceable once going”: we took the lead in establishing a
judicial blockchain platform
The Blue Book: It is acknowledged that the Hangzhou Internet Court has built and
continuously improved “six platforms”—online litigation platform, online mediation
platform, electronic evidence platform, electronic service platform, online execution
platform, trial platform of big data. The whole chain solves conventional judicial
difficulties in case filing, evidence authentication, execution and so on. How do you
think of it?
Du Qian: The Hangzhou Internet Court deeply integrates cutting-edge technolo-
gies to promote the modernization of trial systems and trial capacity. It has upgraded
Internet justice from the Internet to intelligent through “six platforms”. In particular,
the in-depth application of China’s first judicial blockchain platform, a powerful
governance platform that can “discover traces left when going online” from the
source, empowers the source of governance and promotes the standardization of
online behavior with the idea of co-construction, application and governance. In
cooperation with the judicial authorities of Shanghai, Jiangsu and Anhui, the Judi-
cial Blockchain Alliance spanning the Yangtze River Delta region was established to
create a judicial-level trust mechanism with “records of full process, credibility of full
chain and full node witnessing”, promoting the integration of judiciary in the Yangtze
River Delta region. As of April 30, 2020, the platform has over 4 billion pieces of
data on the chain, second only to the “Unified Judicial Blockchain Platform”. On
this basis, the first blockchain smart contract judicial application was launched in
China. Signing, performance and breach of contract, and interpellation are recorded
on the judicial chain in real time, and the entire transaction chain is automatically
documented and enforced. Relying on an AI intelligent judge assistant and intelligent
evidence analysis system, a “full-process online intelligent trial” mechanism is built
300 Y. Cui

to protect the legitimate rights and interests of performing parties at low cost and
high efficiency.
Hearing “Peppa Pig Case”: We took the lead in exploring the Internet judicial
mode across the globe
The Blue Book: In 2018, a case attracted great attention from UK media: Peppa
Pig toys labeled genuine and sold on the Internet were in fact pirated. The British
copyright owner filed a copyright infringement lawsuit of Peppa Pig to Hangzhou
Internet Court. Can you tell us more about this?
Du Qian: The case of copyright infringement of Peppa Pig was featured in The
Times and was praised as “a pioneer to explore Internet judicial mode across the
globe”. It was also included in the Report on the Work of the Supreme People’s Court
and “Top Ten Civil and Administrative Cases of the Year”. The Hangzhou Internet
Court innovated the cross-border judicial coordination mechanism and contributed
to the construction of a community with a shared future for cyberspace. On the
one hand, the Court handles foreign-related disputes in a high-quality way, accumu-
lating vast practical experience and high international credibility in settling disputes
concerning cross-border e-commerce transactions, foreign-related intellectual prop-
erty rights and ownership of international domain names. In addition to the Peppa Pig
case, we closed the first lawsuit of international Internet domain name infringement
brought against the ruling of the World Intellectual Property Organization (WIPO)
and played a leading role in exploring rules and adjudication standards for judicial
protection of international domain names. The Court held that the plaintiff’s inten-
tional registration of the disputed domain name has infringed prior right of Banco
BPM Società per Azioni, and thus dismissing the lawsuit was dismissed in accor-
dance with the law to equally protect the legitimate rights and interests of Chinese and
foreign parties. On the other hand, as the Court takes responsibility for the origin of
Internet justice, it actively demonstrated positive results of China’s Internet judicial
reform to the outside world. The first cross-border trade court was set up to centralize
corresponding cross-border e-commerce and Internet intellectual property disputes
into its jurisdiction, building Hangzhou into a cross-border trade dispute resolution
center located in Asia and facing the world. For two consecutive years, we were
invited to the World Internet Conference and greeted delegations of SCO member
states. On the sixth National Constitution Day, the Court hosted a public open day
of the Supreme People’s Court and welcomed a group of foreign visitors consisting
of presidents of the Supreme Court and heads of the Department of Justice from
over 30 countries and regions, including Greece, South Africa, South Korea, Russia,
Venezuela and Singapore. It provided the international community with a “Chinese
model” of internet governance.
We responded to technological development with judicial innovation and
enhanced judicial productivity.
The Blue Book: The Hangzhou Internet Court is a specialized court closely
connected with information technology. How does the Court build its technological
platform?
7 Expert Interviews on the Development of Artificial Intelligence … 301

Du Qian: As a major institutional innovation of judicial initiative to adapt to the

general trend of Internet development, the Hangzhou Internet Court is empowered by
information technology. It is necessary to react to technological developments with
judicial innovations, transform and apply advanced technologies to judicial projects,
and further enhance judicial productivity. For example, in the domain of judicial
blockchain smart contracts, we have incorporated administrative authorities, judicial
authorities, notary public offices, appraisal organs, scientific institutions, internet
enterprises and evidence preservation platforms into the development, attribution
and governance, and intelligent supervision system of smart contracts, creating a
benign and orderly smart contract ecosystem. At the same time, the application of
judicial technology must comply with technical ethics and morality and be achieved
with a legal and orderly mechanism. We have made efforts to uphold the principle
of technology neutrality, set up an information technology center, recruit specialists
of information technology, take control of the “six platforms”, and complete the
planning, management and implementation of information technology projects.
The Blue Book: The Internet Court needs interdisciplinary talents who are profi-
cient in business and Internet technology applications. How does the Hangzhou
Internet Court, transformed from a railway court, train technical talents and cultivate
judges with AI literacy.
Du Qian: Speaking of staff structure, most personnel in the Internet Court are
transferred from the Railway Transport Court or selected from conventional courts.
In the new trial mode, judges must work on cases and explore how to quickly adapt to
the cross-industry integration of “law + Internet”, fuse blockchain, big data and other
cutting-edge information technologies into judiciary, and use the Internet to resolve
online disputes. The Internet Court requires judges to have a higher command of
knowledge and comprehensive abilities. Fortunately, our team is good at learning and
makes it regular to research new trends of Internet development through organized
studies, academic seminars, research and training. In addition, we have also managed
to attract and retain innovative trial talent through a mechanism of building innovative
teams. We have set up a long-term development plan and institutional design to
help introduce talent and have adopted various methods, such as “four-level” joint
examination, selection of graduates in double first-class universities and employing
civil servants through contracts to attract a number of interdisciplinary talent with
law expertise and Internet technology, to form a pattern in which specialized teams
hear specialized cases.
In addition to knowledge structure and business literacy, it is also important to
transform legal theories and to shape forward-looking ideas and broad perspectives.
The digital economy, based on e-commerce, cloud computing and AI, has become
a new round of industrial revolution since the Industrial Revolution. Emerging will
inevitably bring new issues and massive risks to the digital economy. Only by keeping
up with the Internet mindset can we effectively administer web communities from
the judicial dimension and further extend the value of the rule of law. In this regard,
we will combine going out with bringing in, build a broad platform for officials and
police officers to learn, practice and innovate, encourage them to do more research,
think more diligently, strengthen global governance of cyberspace to make useful
302 Y. Cui

exploration, further enhance the international vision of officials and police officers,
and leave enough room for innovation of judges’ development.
The Blue Book: With the continuous progress of the information society, the
integration of the Internet and society will continue to deepen, and the demand for
the rule of law will also continue to increase. What is your vision of the future
development of the Internet Court?
Du Qian: The Hangzhou Internet Court will continue its reform and innovation,
deepen the mechanism of the judicial Internet and promote the rule of law and
modernization of the cyber governance system and governance capacity.
First, we will continuously boost the deep transformation of the trial mechanism
with a reform mentality and further improve Internet finance, intellectual property
and administration trial mechanisms to provide more experience and samples for
court trials, government administration and social governance.
Second, we will thoroughly explore rules of Internet social governance, clarify
boundaries of exercising rights in the new economy by properly handling new types
of disputes, including disputes involving the Internet of Things, and form a sound
and credible mode of governance of network ecology. Efforts should be made to
promote the substantive operation of the Internet Nomocracy Institute (Hangzhou),
release visual reports on the judicial application of blockchain smart contracts and
“3-in-1” intellectual property trials, provide judicial solutions to society and advance
the good functioning of online social orders.
Third, on the basis of entering a new era of “5G + blockchain” Internet-related
enforcement, we will promote smart enforcement, deepen the reform of the Internet
enforcement mechanism, strengthen the construction of a social credit system, and
provide strong judicial services and guarantees to create a stable, fair, transparent
and predictable business environment.
Fourth, we will follow the direction of technological evolution, give full play to the
first-mover advantage of judicial blockchain research and development and applica-
tion, improve and optimize AI, bravely probe the integration of emerging technolo-
gies such as AI with judicial practice, boost the quality, efficiency and driving force
in judicial adjudication, and systematically enhance the comprehensive governance
capacity of cyberspace.
Fifth, taking the opportunity to promote the construction of cross-border trade
courts with high quality, we will continue to enhance our ability to control the devel-
opment of information technology and ensure cybersecurity through justice, enhance
our international discourse and rule-making power in cyberspace, explore and set up
Internet judicial rules and legal systems with Chinese characteristics, and strive to
contribute more original wisdom to the reform of the global Internet governance
system and the building of a community with a shared future for cyberspace.
Chapter 8
Questionnaire on Legal Aspects
of Artificial Intelligence (with Data
Analysis)

Yadong Cui

1 Questionnaire on Legal Aspects of Artificial Intelligence

In the process of ever deepening integration of the Internet, big data, artificial intel-
ligence and real economy, the legal system may play two roles: it may hinder inte-
gration, or it may guarantee and promote integration. Therefore, on the one hand,
it is necessary to sort out the existing laws and regulations to determine and revoke
the negative laws and regulations that impede deep integration. On the other hand,
relevant laws and regulations need to be introduced and enforced to advance inte-
gration. The realization of these two goals is based on the in-depth study of relevant
legal issues by scholars and practitioners
This questionnaire is designed to gain a more detailed, specific and in-depth under-
standing of legal aspects concerning AI. Your participation is greatly appreciated. All
information provided will be kept confidentially and used solely for testing purposes.
Shanghai Law Society
Yangtze Delta Region Institute of Tsinghua University, Zhejiang
School of Artificial Intelligence and Law, Shanghai University of Political Science
and Law
April 2020
(The questionnaire is posted electronically, and only questions are provided
below.)
Q1–Q4. Information about occupation, age, workplace, types of employer, hierarchy,
etc.
Q5. Do you know any policies or legal documents related to AI development
introduced in recent years in China or other countries?
Q6. Have you followed the Rule of Law Forum of the 2nd WAIC held in Shanghai
for two consecutive years (2018–2019)?

Y. Cui (B)
Shanghai Law Society, Shanghai, China
e-mail: [email protected]

© Shanghai People’s Publishing House 2022 303

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2_8
304 Y. Cui

Q7. What do you think of the practical effects brought by the existing AI-related
policies and legal documents in China?
Q8. Which aspects should be highlighted in the existing AI-related policies and legal
documents?
Q9. In what way do you think should the legislation on AI be improved?
Q10. What should be the top priority of AI policies in China?
Q11. What is your attitude toward the development of AI?
Q12. Has the application of AI in the judicial field in your region improved in the
previous year?
Q13. Do you think the law should be generally open or cautious about the
development of AI? (such as autopilot, facial recognition, smart healthcare)?
Q14. What AI applications have you used?
Q15. Is it necessary to introduce laws or administrative regulations to advance AI
development at the national level?
Q16. What is your impression of AI applications?
Q17–21. What AI technologies (such as facial recognition, etc.) have you used in
drafting and legislative evaluation? Are there any problems with these technologies?
What are the major application scenarios of AI in legislation? What do you think are
the problems with AI application in current legislation in your region? How should
AI be applied in legislation?
Q22. What do you think are the current problems with AI in legal services at the
national level?
Q23. What do you think can be done to build a better platform for intelligent legal
services?
Q24. What AI law enforcement applications have you used?
Q25. Have you encountered privacy leaks due to the use of AI technology (such as
body cams, electronic police, facial recognition and drones) in law enforcement?
Q26. What problems have you encountered in applying AI in law enforcement?
Q27. Compared with natural person law enforcement, what do you think are the
advantages or disadvantages of artificial intelligence robot law enforcement?
Q28. Are there any bugs in the AI applications you have used in the judicial field?
Q29–30. Have you ever used or developed any smart judicial trail systems? What
kind of trials do you use the smart trial system for?
Q31. What are the pros and cons of AI–assisted trial systems?
Q32. What do you think is the future trend of intelligent trial systems?
Q33. How do you think we should use AI to build judicial database?
Q34. Which of the following aspects do you think AI should focus on in the judicial
field?
Q35. What kind of AI projects would you like to develop or introduce in the judicial
field?
Q36. What problems have you encountered when using AI systems, projects or
hardware in your work? What other AI applications do you think are needed in the
judicial field?
Q37. What aspects do you think the Blue Book should focus on?
8 Questionnaire on Legal Aspects of Artificial … 305

2 Data Analysis of the Survey on the Legal Aspects

of Artificial Intelligence

The survey on the Legal Aspects of Artificial Intelligence was conducted from April
5th to May 10th, 2020. The questionnaires were posted on the official public account
of Shanghai Law Society by WeChat, web address sharing and so on. A total of 1444
valid questionnaires were received in the Chinese data collecting system WJX.cn. as
of May 10th, 2020 (Diagrams 8.1 and 8.2).

Diagram 8.1 Source of

audience

Collecting source Number Percentage (%)

WeChat 1436 99.45
Mobile phone 7 0.48
URL 0.07

Province Number Percentage (%) Province Number Percentage (%)

Shanghai 1155 79.88 Tianjin 17 1.18
Anhui 27 1.87 Hebei 12 0.83
Henan 27 1.87 Hubei 10 0.69
(continued)
306 Y. Cui

(continued)
Province Number Percentage (%) Province Number Percentage (%)
Shandong 24 1.66 Zhejiang 10 0.69
Beijing 22 1.52 Sichuan 9 0.62
Jiangsu 21 1.45 Guizhou 9 0.62
Guangdong 18 1.24 Jiangxi 7 0.48
Chongqing 18 1.24 Shanxi 7 0.48

Province Number Percentage (%) Province Number Percentage (%)

Overseas 6 0.41 Yunnan 3 0.21
Gansu 6 0.41 Shanxi 3 0.21
Fujian 5 0.35 Qinghai 2 0.14
Liaoning 5 0.35 Jilin 2 0.14
Hunan 5 0.35 Neimenggu 2 0.14
Xinjiang 5 0.35 Ningxia 1 0.07
Heilongjiang 4 0.28 Hainan 1 0.07
Guangxi 3 0.21

Diagram 8.2 Distribution of audience’s province

In this survey, the data were mainly collected through WeChat, and the audience
was mainly from Shanghai.
Q1. What is your occupation? [Multiple Choice]
8 Questionnaire on Legal Aspects of Artificial … 307

Option Number Percentage (%)

Judge/the prosecutor (including side judge) 335 23. 20
Police 176 12. 19
Administrative law enforcement officials 158 10. 94
Lawyer/legal counsel 167 11.57
AI-related technicians 8 0. 55
Law school students 190 13. 16
Professors/researchers 79 5. 47
AI-related major students 16 1. 11
Others (please be indicated) 310 21. 47
Undetermined 5 0. 35
Valid polls 1444

The option “Others” includes 41 Civil Servants, 36 Social Workers, 26 Side

Judges, 11 People’s Mediators and 8 Prosecution Clerks (Diagram 8.3).

Diagram 8.3 Distribution of audience’s occupations

Q2. How old are you? [Multiple Choice] (Diagrams 8.4 and 8.5)
308 Y. Cui

Option Number Percentage (%)

Below 20 43 2.98
21–30 450 31.16
31–40 571 39.54
41–50 241 16.69
51–60 124 8.59
60 and above 9 0.62
Undetermined 6 0.42
Valid polls 1444

Diagram 8.4 Distribution of audience’s ages

X/Y Judge/the Police Adminis- Lawyer/ AI-related Law Profes- AI- Others Undeter- Valid
prosecutor (%) trative legal techni- school sors/ related (%) mined polls
(including law counsel cians students researchers major (%)
side enforce- (%) (%) (%) students
judge) (%) ment (%)
officials
(%)

Below 20 0.00 0.00 0.00 0.00 0.00 74.42 0.00 18.60 6.98 0.00 43
21–30 25.06 4.21 4.88 9.76 0.44 34.81 1.55 1.33 17.96 0.00 451
31–40 24.48 16.96 13.64 12.06 0.17 0.35 4.02 0.17 28.15 0.00 572
41–50 20.33 16.18 15.35 16.60 1.24 0.00 12.45 0.41 17.43 0.00 241
51–60 25.81 17.74 16.94 8.06 1.61 0.00 14.52 0.00 15.32 0.00 124
60 and 11.11 0.00 0.00 44.44 0.00 0.00 11.11 0.00 33.33 0.00 9
above
8 Questionnaire on Legal Aspects of Artificial … 309

Diagram 8.5 The relation between occupation and age

The main body of this survey is students below 20 in higher education. Officials
in the prosecution and judicial system are distributed in a balanced proportion as
representatives of other age groups.
Q3. Where is your workplace? [Gap Filling] (Diagrams 8.6, 8.7 and 8.8)

Diagram 8.6 Distribution of audience’s workplace

310 Y. Cui

Diagram 8.7 Distribution

of audience in Shanghai

Diagram 8.8 Distribution

of audience in Beijing
8 Questionnaire on Legal Aspects of Artificial … 311

Q4. What is the type of your employer and hierarchy? [Gap Filling] (Diagrams
8.9, 8.10, 8.11, 8.12 and 8.13)

The procuratorate 161

Undetermined 53
Court 233
Colleges and universities 227
Law enforcement 165
Others 607

Diagram 8.9 Distribution of employer

Diagram 8.10 Distribution

of police hierarchy
312 Y. Cui

Diagram 8.11 Distribution of judges in different courts

Diagram 8.12 Distribution of the prosecutors

8 Questionnaire on Legal Aspects of Artificial … 313

Diagram 8.13 Distribution of higher education

Q5. Do you know any policies or legal documents related to AI develop-

ment introduced in recent years in China or other countries? [Multiple Choice]
(Diagram 8.14)

Option Number Percentage (%)

I am very concerned about them and have read them before 117 8.10
I am concerned about them and have read related introduction and news of 523 36.22
them
I have heard of them but didn’t know details 573 39.68
I am not familiar with them 221 15.30
Undetermined 10 0.69
Valid polls 1444
314 Y. Cui

Diagram 8.14 The relation between age and attention to AI-related policies and legal regulations

From the perspective of age, the group between 21 and 40 is less concerned
about AI-related policies and legal documents, which indicates that concern is nega-
tively correlated with age. However, the percentage of “very concerned about” was
increased in the 41 and above group (Diagram 8.15).

Diagram 8.15 The relation between occupation and attention to AI-related policies and legal
regulations
8 Questionnaire on Legal Aspects of Artificial … 315

Within the perspective of occupation distribution, except for AI-related tech-

nicians, concern is gradually reducing in the group of university teachers, judges
and prosecutors, administrative law enforcement officials, lawyers and police. What
needs to be stressed is that students who major in AI and law know little about AI-
related policies and legal regulations. In this group, students who chose the option
“I am not familiar with them” reached 12.50%, and the option “not concerned about
them” accounted for 18.75%, which obviously exceeds other groups of audiences.
Based on Diagrams 8.16, 8.17, 8.18 and 8.19, no significant differences are found
in the concern for AI-related policies and legal regulations among the various groups
(Diagram 8.20).

Diagram 8.16 Judges’ attention to artificial intelligence policies and regulations

316 Y. Cui

Diagram 8.17 Prosecutors’ attention to artificial intelligence policies and regulations

Diagram 8.18 The police’s attention to artificial intelligence policies and regulations
8 Questionnaire on Legal Aspects of Artificial … 317

Diagram 8.19 University faculty and researchers’ attention to artificial intelligence policies and
regulations

Urban areas Suburban areas

Diagram 8.20 Comparison of the attention to AI policies between urban and suburban areas in
Shanghai

The diagram indicates that people in urban areas pay more attention to AI than
those who live in suburban areas.
Q6. Have you followed the Rule of Law Forum of 2nd WAIC held in Shanghai
for two consecutive years (2018–2019)? [Multiple Choice] (Diagram 8.21).
318 Y. Cui

Option Number Percentage (%)

I have attended to it and know many details of it 80 5.54
I have a general idea of it through news 382 26.45
I have read the news of it but know little details 629 43.56
I am not familiar with the forum 344 23.85
Undetermined 9 0.62
Valid polls 1444

Diagram 8.21 The relation between age and familiarity with rule of law forum of WAIC

The increase in age is positively correlated with knowledge of the Rule of Law
Forum of the WAIC. However, the option of “I don’t know” still bears the lion’s
share of the proportion. Young students polarize the proportion on the first option
and the fourth option (Diagram 8.22).
8 Questionnaire on Legal Aspects of Artificial … 319

Diagram 8.22 Attention of different occupations to the rule of law forum of the WAIC

The officials of the prosecution and judicature are the main groups that choose the
second and third options, while the students and technicians related to AI are major
contributors to the second option. For students who major in AI, their familiarity with
the Rule of Law Forum of the WAIC is far greater than that of AI-related policies and
legal regulations, which indicates that AI-related majors are less concerned about the
policies and regulations.
Q7. What do you think of the effects of the existing AI-related policies and legal
documents in China? [Multiple Choice] (Diagram 8.23)

Option Number Percentage (%)

Excellent 168 11.63
Good 214 14.82
Somehow useful 484 33.52
Insignificant 149 10.32
Not sure 421 29.16
Undetermined 8 0.55
Valid polls 1444
320 Y. Cui

Diagram 8.23 The relation between policies acceptability and audience occupation

For the groups of AI-related technicians and students, there is a polarization

between the options of “Insignificant” and “Excellent”. For law school students,
they deem that the practical effect of policies and regulations related to AI is lower
than the expectation. This attitude is reflected in the survey, especially through the
low proportion of the option “good” chosen by them (Diagram 8.24).

Diagram 8.24 The relation between attention to AI-related policies and regulations and the
acceptability of its practical effect
8 Questionnaire on Legal Aspects of Artificial … 321

The attention and knowledge of AI-related policies and regulations is positively

correlated with the acceptability of its effect (Diagram 8.25).

Diagram 8.25 The relation between the familiarity with and acceptability to the rule of law forum
of WAIC

The more you know about the Rule of Law Forum of WAIC, a better recognition
will be received to the policies and regulations. This positive correlation is explicitly
shown by the high proportion of the option “Excellent”.
Q8. Which aspects should be highlighted in the existing AI-related policies and
legal documents in China? [Multiple Choice (you can choose more than one option
in sequence)] (Diagrams 8.26 and 8.27)

Option Accumulated average score

Risk control and legal liability 4.91
Opening data in an orderly and legally manner 3.81
A unifiable organization for the establishment of polices and laws 3.61
Ethics of design and use 3.12
Promotion and guarantee of basic scientific research 3.09
Strengthen the top-level design of the system 3.04
International communication and cooperation 2.14
Other(s) 0.32
322 Y. Cui

Diagram 8.26 The first essential concerns

Diagram 8.27 The second essential concerns

The options of “Risk Control and Legal Liability” and “Opening Data in an
Orderly and Legally Manner” are two primary concerns of the public in the current
AI-related policies and legal documents (Diagrams 8.28, 8.29, 8.30 and 8.31).
8 Questionnaire on Legal Aspects of Artificial … 323

Diagram 8.28 Distribution of the options by judges and the prosecutors

Diagram 8.29 Distribution of the options by police

324 Y. Cui

Diagram 8.30 Distribution of the options by administrative officials

Diagram 8.31 Distribution of the options by teachers and researchers

8 Questionnaire on Legal Aspects of Artificial … 325

Based on the diagrams above, teachers and researchers pay the most attention
to risk control and legal liability, while police and administrative officials are more
concerned about the “Strengthening the Top-Level Design of the System”.
Q9. In what way do you think should the legislation on artificial intelligence be
improved? [Please arrange your choices in order.] (Diagrams 8.32, 8.33 and 8.34)

Option Accumulated average score

Encourage qualified companies to carry out research on the laws 2.17
and policies regulating AI
Bringing the rule of law of AI into the schedule of relevant 2.12
departments
Research on the AI and rule of law supported by the national 2.02
scientific research fund
Strengthen international cooperation to understand the 1.59
international rule of law

Diagram 8.32 Distribution of the options by judges and the prosecutors

326 Y. Cui

Diagram 8.33 Distribution of the options by police

Diagram 8.34 Distribution of the options by teachers and researchers

8 Questionnaire on Legal Aspects of Artificial … 327

Based on the diagrams above, judges and prosecutors choose “to encourage qual-
ified companies to carry out research on laws and policies regulating AI. Police “to
formally incorporate the laws and policies regulating AI into the work plan of rele-
vant departments”, while university teachers and researchers choose the “Research
on AI and rule of law supported by the National Scientific Research Fund”.
Q10. What should be the top priority of AI policies in China now? [Multiple
Choice] (Diagrams 8.35 and 8.36)

Option Number Percentage (%)

Giving priority in supporting leading enterprises so that they could 428 29.64
drive the industrial chain(s)
Giving priority to create better business environments for startups so 681 47.16
that they can compete and emerge as leaders
Market entities should be treated equally 321 22.23
Undetermined 14 0.97
Valid polls 1444

Diagram 8.35 The top priority of AI policies in China

328 Y. Cui

Diagram 8.36 Relation between occupations and preference of policies

Judges and prosecutors focus more on giving priority to creating better business
environments for start-ups so that they can compete and emerge as leaders, while
police think priority should be given to supporting leading enterprises so that they
can drive the industrial chain(s) (Diagram 8.37).
8 Questionnaire on Legal Aspects of Artificial … 329

Diagram 8.37 Relation between concern with AI-related policies and priority in policies design

There is a positive correlation between the concern of policies and the way of
giving priority to supporting leading enterprises so that they can drive the industrial
chain(s).
Q11. What is your attitude toward the development of AI? [Multiple Choice]
(Diagrams 8.38 and 8.39)

Option Number Percentage (%)

AI will benefit humanity 353 24.45
Relative positive. AI will benefit humanity overall, however it might 897 62.12
lead to new social problems
Relative negative. AI will bring uncertainties to the society, but it is 56 3.88
beneficial for local advancement
AI will detract human subjectivity, and lead to inequity. It should be 23 1.59
strictly limited
It is hard to say 101 6.99
Undetermined 14 0.97
Valid polls 1444
330 Y. Cui

Diagram 8.38 Attitude toward AI in different age groups

Diagram 8.39 Attitudes of different occupation groups toward AI

8 Questionnaire on Legal Aspects of Artificial … 331

From the age groups of “below 20”–60, with the increase in age, the number of
people who have absolutely positive attitudes toward AI continues to rise sharply.
The number of people who held a relatively positive attitude continued to decline
significantly, while the proportion of people with negative and partial negative atti-
tudes remained stable. The age group of 60 and above embraces the most positive
attitude toward AI (Diagram 8.40).

Diagram 8.40 The relation between attention and attitude to AI-related policies and regulations

A positive correlation is found between attention to AI-related policies and

regulations and attitudes toward AI (Diagram 8.41).
332 Y. Cui

Diagram 8.41 The relation between the publicity of rule of law forum and attitude toward AI

A positive correlation is found between the publicity of the Rule of Law Forum
and attitudes toward AI.
Q12. Has the application of AI in the judicial field in your region improved in the
previous year? [Multiple Choice]

Option Number Percentage

Visible improvements 573 39.68%
No improvements 199 13.78%
Not sure 655 45.36%
Undetermined 17 1.18
Valid polls 1444

The audience who chose the option “visible improvement” mentioned the 206
AI-assisted case handling system adopted by Shanghai Courts. In addition, other
online judicature applications, such as the WeChat Mobile Mini Court, used to place
a case on file, online trail and AI robot named Hu Xiaofa, meet the demand of timely
prefecture-level legal counsel. By using these applications, online judicial working
efficiency has been greatly improved, and it is explicitly shown in the following
examples: online trail is more convenient and more stable against the backdrop of
COVID-19, the speed of inquiring cases and articles is accelerated, commutation and
parole can be handled online, the use of body cam is regulated, a smart command
center is built, and the automatic legal files identification system adopted by Hong
Kong Court, Shanghai to arrange the documents, which greatly improve the efficiency
in filing cases (Diagram 8.42).
8 Questionnaire on Legal Aspects of Artificial … 333

Diagram 8.42 Distribution of options among different occupations

The attitudes of AI-related technicians are relatively negative.

Q13. Do you think the law should be generally open or cautious about the devel-
opment of artificial intelligence? (such as autopilot, facial recognition, smart health
care)? [Multiple Choice]

Option Number Percentage (%)

Margin of error should be given to AI 144 9.97
Relative open, the development of the future cannot be limited by the 463 32.06
laws from the traditional society, while some necessary bottom lines
need to be preserved
Relative cautious, laws should be used to sustain the value and order 557 38.57
of currently, while it could be opened in some specific areas
Laws are the bottom line of the human society, they cannot be 180 12.47
released unless the AI technology is fully matured within the lab
Not sure 83 5.75
Undetermined 17 1.18
Valid polls 1444

The attitudes of AI-related technicians toward AI-related law are polarized, with
both open and cautious attitudes having an overall majority, especially the people
who embrace cautious attitudes accounting for 25%. There is no significant effect
on other professions. The greater they understand and pay attention to policies and
regulations, the more open their attitude is forging toward the law (Diagrams 8.43,
8.44 and 8.45).
334 Y. Cui

Diagram 8.43 Attitudes toward AI applied to legal aspects in different occupations

Diagram 8.44 The relation between the familiarity with the rule of law forum of WAIC and attitude
toward AI applied in legal aspects
8 Questionnaire on Legal Aspects of Artificial … 335

Diagram 8.45 The relation between the concern to AI-related polices and regulations and attitude
toward AI applied in legal aspects

Q14. What AI applications have you used? [Multiple Choice (You can choose
more than one option)] (Diagram 8.46)

Option Number Percentage (%)

Execution and supervision of law 632 43.77
Detection and processing of traffic violations 977 67.66
Patrol interrogation and examination 362 25.07
Legal advice 581 40.24
Assist in handling cases 652 45.15
Others 45 3.12
Not sure 92 6.37
Non valid polls 14 0.97
Valid polls 1444
336 Y. Cui

Diagram 8.46 Distribution of AI technologies exposure to different occupations

Q15. Is it necessary to introduce laws or administrative regulations to advance AI

development at the national level? [Multiple Choice] (Diagram 8.47)

Policies have an urgent need for specific laws regarding AI (Diagram 8.48).
8 Questionnaire on Legal Aspects of Artificial … 337

Diagram 8.47 Distribution of options by occupations

Diagram 8.48 Attention and attitude to AI policies and regulations

338 Y. Cui

The more attention you pay to the policies and regulations, the more support it
would receive in introducing policies and regulations regarding AI.
Q16. What is your impression of AI applications? [Arrange your options in order]
(Diagrams 8.49, 8.50, 8.51 and 8.52)

Option Accumulated average score

Convenient and efficient 5.25
Convenient electronic forensics 3.34
Open and transparent procedure 2.88
Crack down on illegal activities 2.25
Pose pressure and burden 0.82
No feelings 0.55
Not sure 0.13

Diagram 8.49 Impression of AI applications by judges and the prosecutors

8 Questionnaire on Legal Aspects of Artificial … 339

Diagram 8.50 Impression of AI applications by police

Diagram 8.51 Impression of AI applications by administrative officials

340 Y. Cui

Diagram 8.52 Impression of AI applications by law school students

Q17. Are you engaged in legislative work? [Multiple Choice]

Option Number Percentage (%)

YES 38 2.63
NO 1384 95.84
Undetermined 22 1.52
Valid polls 1444

Q18. What AI technologies (such as facial recognition, etc.) have you used in
drafting and legislative evaluation? Are there any problems with these technologies?
[Gap Filling]
Q19. What are the major application scenarios of AI in legislation? [Multiple
Choice (with more than one options)]

Option Number Percentage (%)

Soliciting and collecting opinions 24 63.16
Normative documents filing 17 44.74
Data retrieval and reference 25 65.79
Others 3 7.89
Undetermined 1 2.63
Valid polls 38

Q20. What are the problems with AI applications in legislation in your region?
[Multiple Choice (with more than one options)]
8 Questionnaire on Legal Aspects of Artificial … 341

Option Number Percentage (%)

Massive data lacks arrangement; thesaurus is disorganized and lacks management 16 42.11
Regional differences caused by local economic strength, talents and technology 19 50
Inactive attitude of the local governments 10 26.32
Lack of talents who know how to apply AI legal fields 21 55.26
Lack of funds 11 28.95
Lack of technologies 20 52.63
Lack of local legislative environment and legislative demand 14 36.84
Others 2 5.26
Undetermined 1 2.63
Valid polls 38

Q21. How should AI be applied to legislation?[Multiple Choice]

Option Number Percentage (%)

Only appear in the form of databases. It provides information retrieval, opinion 17 44.74
collection, and text checking for legislative work
Utilize the algorithmic solving function of AI. In addition to providing 20 52.63
information, it is allowed to participate in the drafting and legislative evaluation
Others 0 0
Undetermined 1 2.63
Valid polls 38

Q22. What do you think are the current problems with AI in legal services at the
national level? [Multiple Choice (with more than one option)] (Diagram 8.53)

Option Number Percentage (%)

The range of services are not diversified. It mainly provides online consultation, 750 51.94
search of judgment documents and lawyer matching service
Service quality needs to be improved. It cannot provide timely and efficient 831 57.55
solutions to legal problems
Network and software design issues 621 43.01
Unnecessary to use 100 6.93
Over commercialization 225 15.58
Troublesome operation 266 18.42
Concern of privacy protection and information security 636 44.04
No policies and regulations regarding AI legal services 448 31.02
Others 29 2.01
Undetermined 20 1.39
Valid polls 1444
342 Y. Cui

Diagram 8.53 Distribution of occupations of Q22

Q23. What can we do to build a better AI legal service platform? [Multiple Choice
(with more than one option)] (Diagram 8.54)

Options Number Percentage

To publicize the smart legal service through Apps and encourage the competition 923 63.92%
among developers
To enrich the range of legal services, such as legal search, contract drafting, document 1139 78.88%
review, case forecasting, etc
Use reward mechanisms, ranking rewards, and quota grabbing to improve the response 405 28.05%
efficiency and answer quality of questions
Other (s) 23 1.59%
Conduct the research and build the public welfare platforms with the leading of 427 29.57
government
To build the platforms with the helps of court 335 23.2%
Undetermined 22 1.52%
Valid polls 1444
8 Questionnaire on Legal Aspects of Artificial … 343

Diagram 8.54 Distribution of occupations in this Q23

Q24. What AI law enforcement applications have you used? [Matrix Multiple
Choice] (Diagrams 8.55 and 8.56)

Option I have used it I have heard of it Undetermined

Body cam 322 (22.3%) 515 (35.66%) 634 (43.91%)
PDA mobile devices for policing 161 (11.15%) 225 (15.58%) 1066 (73.82%)
Electronic police (Camera) 213 (14.75%) 719 (49.79%) 531 (36.77%)
Intelligent traffic monitoring and logging 145 (10.04%) 420 (29.09%) 890 (61.63%)
system
Community police-end app 101 (6.99%) 220 (15.24%) 1129 (78.19%)
Drones 115 (7.96%) 358 (24.79%) 985 (68.21%)
344 Y. Cui

Diagram 8.55 Distribution of judges and the prosecutors’ choices

Diagram 8.56 Distribution of police’s choices

Polices are the main users of AI applications applied in law enforcement

(Diagram 8.57).
8 Questionnaire on Legal Aspects of Artificial … 345

Diagram 8.57 Distribution of administration officials’ choices

Q25. Have you encountered privacy leakage due to the use of AI technology (such
as body cam, electronic police, facial recognition and drones) in law enforcement?
[Multiple Choice] (Diagram 8.58)

Option Number Percentage (%)

Yes 114 7.89
No 357 24.72
Not sure 373 25.83
Worried 578 40.04
Undetermined 22 1.52
Valid polls 1444

X/Y Yes No Not clear Concerned Undetermined Number

Judge/the 21 (6.27%) 82 (24.48%) 102 (30.45%) 125 (37.31%) 5 (1.49%) 335
prosecutor
(including side
judge)
Police 16 (9.04%) 72 (40.68%) 42 (23.73%) 45 (25.42%) 2 (1.13%) 177
administrative 16 (10.13%) 46(29.11%) 34 (21.52%) 59 (37.34%) 3 (1.90%) 158
law
enforcement
officials
Lawyer/legal 16 (9.58%) 30 (17.96%) 36 (21.56%) 83 (49.70%) 2 (1.20%) 167
counsel
AI-related 2 (25%) 3 (37.5%) 2 (25%) 1 (12.5%) 0 (0.00%) 8
technicians
(continued)
346 Y. Cui

(continued)
X/Y Yes No Not clear Concerned Undetermined Number
Law school 15 (7.85%) 26 (13.61%) 50 (26.18%) 96 (50.26%) 4 (2.09%) 191
students
Professors/ 14 (17.72%) 14 (17.72%) 17 (21.52%) 34 (43.04%) 0 (0.00%) 79
researchers
AI-related 0 (0.00%) 1(6.25%) 4 (25%) 10 (62.5%) 1 (6.25%) 16
major students
Others (please 14 (4.52%) 83 (26.77%) 86 (27.74%) 126 (40.65%) 1 (0.32%) 310
be indicated)

Diagram 8.58 Distribution of occupations in Q25

Polices are less concerned about privacy leakage, while lawyers, law school
students and AI-related major students are very worried about it.
Q26. What problems have you encountered in applying AI in law enforcement?
[Multiple Choice (with more than one option)]

Option Number Percentage (%)

Operational issues. It’s difficult for law enforcement officials to learn 562 38.92
the operation of AI
Privacy security AI tends to invade individual privacy 1032 71.47
Technical issues. AI devices are greatly affected by weather, light 664 45.98
and various factors
Data barriers. Insufficient data interconnection 791 54.78
Others 35 2.42
Undetermined 20 1.39
Valid polls 1444
8 Questionnaire on Legal Aspects of Artificial … 347

The option of “Others” includes: the lack of professional analysis after collecting
data; the application of AI is immature, the necessity of applying AI devices; the
complexity of the report and approval procedure, especially the problems existed
in the connection of AI devices; for the ground that AI is not as smart as human
being, there is much room for improving the data quality and the data collection
(Diagram 8.59).

Diagram 8.59 Distribution of occupations in Q26

Police concerns most operational issues, while teachers and researchers focus
more on privacy security.
Q27. Compared with natural person law enforcement, what do you think are
the advantages or disadvantages of artificial intelligence robot law enforcement?
[Multiple Choice (with more than one option)]

Option Number Percentage (%)

AI to prevent malpractice for favoritism 862 59.7
Concern for privacy invasion by AI 784 54.29
AI can save both time and labor 837 57.96
AI will cause unexpected error in law enforcement due to technical 756 52.35
problems
Others 33 2.29
Undetermined 25 1.73
Valid polls 1444
348 Y. Cui

The option of “Others” includes:

Whether AI can handle different situations in law enforcement. When there is no
matching option, AI may forcibly select an illegal act. AI may change background
data. The basis of AI learning is real data, and it is worried that statistical fraud will
lead to false results in AI learning. AI cannot handle emotional factors involved in
cases, which can be seen as an advantage or a disadvantage in different circumstances.
We should pay attention to striking a balance between the rule of humans, rule of
law and rule of AI. Technical barriers may lower the efficiency of AI applied in law
enforcement. It is worried that AI law enforcement will inevitably be inflexible and
rigid (Diagram 8.60).

X\Y A1 to Concern Al can save AL will Others Undetermined Valid

prevent for privacy both time cause polls
malpractice invasion by and labor unexpected
for Al error in law
favoritism enforcement
due to
technical
problems
Judge/the 185 178 189 175 ( 8 4 (1.19%) 335
prosecutor (−55.22%) (−53.13%) (56.42%) 52.24%) (2.39%)
(including side
judge)
Police 105 80 107 82 1 3 1 (−0.56%) 177
(59.32%) (45.20%) (60.45%) 46.33%) (−1.69%)
Administrative 98 90 89 71 (44.94%) 2 4 (−2.53%) 158
law (−62.03%) (−56.96%) (−56.33%) (−1.27%)
enforcement
officials
Lawyer/legal 107 90 99 88 4 3 (1.80%) 167
counsel (64.07%) (−53.89%) (59.28%) (−52.69%) (−2.40%)
Al-rclatcd 5 (62.5%) 2 (25%) 4 (−50%) 5 (62.5%) 0 0 (0.00%) 8
technicians (−0.0(1%)
Law school 103 124 116 130 2 6 (3.14%) 191
students (−53.93%) (−64.92%) (−60.73%) (−68.06%) (−1.05%)
Professors/ 49 57 50 36 0 0 (0.00%) 79
researchers (−62.03%) (72.15%) (63.29%) (−45.57%) (0.00%)
AI-related 13 6 9 9 1 1 (−6.25%) 16
major students (81.25%) (−37.5%) (56.25%) (−56.25%) (6.25%,)
Others 198 158 174 161 13 2 (0.65%) 310
(undetermined) (63.87%) (50.97%) (−56.13%) (−51.94%) (4.19%)
8 Questionnaire on Legal Aspects of Artificial … 349

Diagram 8.60 Distribution of occupations in Q27

Q28. Are there any bugs in the AI applications you have used in the judicial field?
[Multiple Choice] (Diagram 8.61)

Option Number Percentage (%)

Often appear 133 9.21
Occasionally appear 554 38.37
Never appear 674 46.68
Undetermined 83 5.75
Valid polls 1444
350 Y. Cui

Diagram 8.61 Distribution of occupations in Q28

Q29. Have you ever used or developed any smart judicial trail systems? [Multiple
Choice (with more than one option)]

Option Number Percentage (%)

WISE system for research and judgment (“Rui Fa Guan”) 28 5.48
AI-assisted case handling system for criminal cases, Shanghai 237 46.38
Intelligent batch cases review system 40 7.83
Others 32 6.26
No use and development 204 39.92
Undetermined 6 1.17
Valid polls 511

X\Y WISE Al-assisted Intelligent Others No use and Undetermined Valid

system case batch development polls
for handling cases r
research system for eview
and criminal system
judgment cases,
(“Rui Fa Shanghai
Gtian”)
Judge/the 19 147 31 27 126 5 (−1.49%) 335
prosecutor (5.67%) (−43.88%) (9.25%) (8.06%) (37.61%)
(including
side
judge)
Police 9 91 9 5 78 1 (0.56%) 177
(−5.08%) (51.41%) (−5.08%) (−2.82%) (−44.07%)
8 Questionnaire on Legal Aspects of Artificial … 351

Q30. What kind of trials do you use the smart trial system for? [Multiple Choice]

Option Number Percentage (%)

Civil and commercial cases 82 27.24
Criminal cases 190 63.12
Administrative cases 16 5.32
Undetermined 13 4.32
Valid polls 301

Q31. What are the pros and cons of AI–assisted trial systems? [Multiple Choice
(with more than one option)]

Option Number Percentage (%)

It works well and there are no major defects 9 4.41
The document assistive function is effective 52 25.49
Digital case files are convenient 95 46.57
Providing intelligent assistant for case reference, evidence, 56 27.45
contradiction, process specification, etc
The materials used during the trial are convenient and intuitive 42 20.59
The level of technological development is relatively low, so that it’s 53 25.98
difficult to achieve technological breakthroughs such as deep learning
The technical standard is not unified and the system is not 57 27.94
compatible, which may results different judgments in different courts
The data sample is not comprehensive, so it’s difficult to support 50 24.51
intelligent analysis
The rules of the algorithm are vague, and the process of judging is not 32 15.69
open and transparent
The text assistance function is not smooth 42 20.59
Others 31 15.2
Undetermined 7 3.43
Valid polls 204

Q32. What is the trend of AI trail systems in the future? [Multiple Choice (with
more than one option)]

Option Number Percentage (%)

It might replace the judge as the adjudicator 32 6.26
It could be used as a clerical aid to hear cases 363 71.04
It could provide prejudgment references for case information 310 60.67
It’s not optimistic to develop the intelligent trial system 46 9
Not sure 55 10.76
(continued)
352 Y. Cui

(continued)
Option Number Percentage (%)
Undetermined 9 1.76
Valid polls 511

Q33. How do you think we should use AI to build judicial database? [Multiple
Choice (with more than one option)]

Option Number Percentage (%)

Using the data of social think-tank for data mining 307 60.08
Building a national unified judicial database platform and breakdown 237 73.39
judicial data barriers
Using AI to identify, summarize and process data 40 69.47
Others 32 2.54
Undetermined 6 1.57
Valid polls 511

Q34. Which of the following aspects do you think AI should focus on in the
judicial field? [Multiple Choice (with more than one option)]

Option Number Percentage (%)

Intelligent trial and smart assistance for judgment in trial 275 53.82
Judicial database development 345 67.51
Court service enforcement 269 52.64
Intelligent legal documents making 318 62.23
Intelligent inspection service 238 46.58
Construction of data sharing service platform for judicial 305 59.69
administration system
Undetermined 6 1.17
Valid polls 511

Q35. What kind of AI projects would you like to develop or introduce in the
judicial field? [Multiple Choice (with more than one option)

Option Number Percentage (%)

AI products designed to release the burden of litigation, such as 326 63.8
judicial and administrative intelligence service terminal
Reduce the simple repetition of work by judicial staff, such as the 433 84.74
AI project in filing
AI hardware that assist in handling cases, such as intelligent 345 67.51
mobile recorders
(continued)
8 Questionnaire on Legal Aspects of Artificial … 353

(continued)
Option Number Percentage (%)
Others 11 2.15
Undetermined 7 1.37
Valid polls 511

Q36. What problems have you encountered when using AI systems, projects or
hardware in your work? What other AI projects do you think are needed in the judicial
field? [Gap Filling]
Some suggestions are as follows:
Although court records are replaced by audio and video recordings or speech
recognition, the court clerks still need to modify the transcripts in the courtroom in
person, which is not as time-saving as expected. Moreover, electronic files are used
to replace paper-based files due to their convenience, but the task of transforming
the paper-based files into electronic files still falls on the clerks. However, the lack of
personnel and judicial resources poses an extra workload to the court, and more time
needs to be spent on them. For example, the Intermediate People’s Court in Shanghai
no longer receives any paper files of appeal cases, and the clerks of the Prefecture
People’s Courts in the same place need to additionally scan the paper files that have
been filed electronically before, some of which are more than 4,000 pages and it is
time-consuming to scan, which undoubtedly poses more burden to the clerks. The
party who files the lawsuit online does not need to submit any paper-based materials,
but the entire process is still time costly because the clerk needs to download and
print out the electronic materials from the computer and mail them to the defendant.
The effectiveness of electronic delivery still needs to be improved, and the preceding
database and article-auxiliary system have yet to be developed. The current problem
that needs to be considered is that judicial personnel are too limited to handle the
increasing number of cases submitted to the Prefecture People’s Court. An auxiliary
system for judgment with elderly friendly functions is expected to develop as soon
as practicable to improve the working efficiency. In addition, speech-to-text systems
applied to court records also require more artificial intelligence technologies, and
they are expected to automatically recognize and summarize key sentences of the
records rather than simply transliterating them.
Q37. What aspects do you think the Blue Book (2020) should focus on? [Gap
Filling]
Among all the questionnaires, 190 do not have any suggestions. For other sugges-
tions submitted with the questionnaires, 41 suggestions concerned privacy protection,
and 16 concerned legal ethics and social ethics.
Appendix A
List of Policies and Regulations on Artificial
Intelligence (2019.7–2020.6)

China
1. Ministry of Science and Technology Notice on the Publication of the Guidelines
for National New Generation Artificial Intelligence Innovation and Development
Pilot Zone Construction Work, (2019.8).
2. Letter from the Ministry of Science and Technology on Supporting Beijing
in Building a National Pilot Zone for the Innovative Development of New
Generation Artificial Intelligence, (2019.10).
3. Letter from the Ministry of Science and Technology on Supporting Shanghai in
Building a National Pilot Zone for the Development of New Generation Artificial
Intelligence, (2019.10).
4. Industrial Structure Adjustment Guidance Catalog (2019), (2019.11).
5. The State Forestry and Grassland Administration, Guidance on Promoting the
Development of Artificial Intelligence in Forestry and Grassland, (2019.11).
6. Ministry of Science and Technology, Notice on the Publication of the Guidelines
for National New Generation Artificial Intelligence Open Innovation Platforms
Construction Work, (2019.12).
7. The Ministry of Education, National Development and Reform Commission and
Ministry of Finance, Notice on the Publication of Several Opinions on Promoting
the Integration of Disciplines and Accelerating the Training of Postgraduates in
Artificial Intelligence in Universities under the Construction of “Double First
Class University Plan”, (2020.1)
8. Letter from the Ministry of Science and Technology on Supporting Chengdu to
Build a National Pilot Zone for Innovation and Development of New Generation
Artificial Intelligence, (2020.1).
United States
1. Navy Artificial Intelligence Framework (2019.9).
2. American AI Century: A Blueprint for Action (2019.12).
3. Artificial Intelligence and National Security (2019.11).
4. The Defense Authorization Act for Fiscal Year 2019 (2019.11).

© Shanghai People’s Publishing House 2022 355

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2
356 Appendix A: List of Policies and Regulations on Artificial Intelligence (2019.7–2020.6)

5. Guidance for Regulation of Artificial Intelligence Applications (2020.1).

6. Bill on Ethical Use of Face Recognition (draft) (2020.2).
European Union
1. Shaping Europe’s Digital Future (2020.2).
2. White Paper on Artificial Intelligence (2020.2).
3. European Data Strategy (2020.2).
4. EU Industrial Strategy (2020.3).

South Korea
National Strategy for Artificial Intelligence (2019.12).
Russia
Russia’s National Al development Strategy by 2030 (2019.10).
Appendix B
List of Artificial Intelligence and Law Research
Institutions (2019.7—2020.6)

(Newly established) Institutions in China

1. The Joint Laboratory of Artificial Intelligence, Criminal Justice School,

Zhongnan University of Economics and Law. The lab was jointly established
by the Criminal Justice School at Zhongnan University of Economics and Law
and Nanjing Xinktech Co., Ltd.
2. The Research Center of Big Data, Artificial Intelligence and Judicial Manage-
ment, Hunan University. The center was jointly established by Hunan University,
Hunan Provincial High People’s Court and Hunan Provincial People’s Procura-
torate, aiming at encouraging a two-way flow of talent between institutions,
exploring a new mode of learning and training, building a practical teaching base
for law, and cooperating in theoretical research.
3. The Research Center of Science and Technology Law, University of Science
and Technology Beijing. The center was established in response to the national
demand for developing science and technology and related legal regulations and
ethical values.
4. The Artificial Intelligence Rule of Law Research Association, Shanghai Law
Society. The association was established on August 19, 2019 and was dedicated
to theoretical and practical research on AI and rule of law, serving as a “think
tank” for promoting the theoretical development of AI and rule of law in China.
5. The Institute of Digital Rule of Law, East China University of Political Science
and Law. The institute was established on June 9, 2020, based on the School of
Law and the Department of Jurisprudence, integrating advance research resources
of the university, and now has a research team of more than 30 people. The
Institute of Digital Rule of Law is headed by Professor Ma Changshan, with
Researcher Lu Yufeng and Associate Researcher Han Xuzhi as Vice Deans.

© Shanghai People’s Publishing House 2022 357

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2
358 Appendix B: List of Artificial Intelligence and Law Research Institutions …

(Newly established) Institutions in Other Countries

1. Ethical Approach to AI” (EAA). The EAA is a committee of external individuals

to discuss AI-related topics from an ethical and legal perspective, and it was
established by ABEJA Inc. on July 30, 2019.
2. VinAI, Japan, an Artificial Intelligence Research Institute Established by
VinTech, a Subsidiary of Vingroup. Vin AI is Used not Only in AI and Machine
Learning but also in Education, Legal, Consulting and Other Related Businesses.
Appendix C
List of Important Publications and Papers
on Artificial Intelligence and Law
(2019.7—2020.6)

Publications and Papers in China

1. Liu Xianquan. (2019, July). Criminal Law in the age of Artificial Intelligence.
Shanghai People’s Publishing House.
2. Sun Zhanli & Sun Zhiwe. (2019 November 3). Study on Frontier Legal Issues
on Artificial Intelligence and Internet. Law Press.
3. Zhao Wanyi & Hou Dongde. (2020, January). The Artificial Intelligence Age of
Law. Law Press.
4. Huayu Yuandian Institute of Legal Artificial Intelligence. (2020, May). A Legal
Professional’s Guide to the Artificial Intelligence. Law Press
5. Gu Jun. (2020, May). The Legal Thinking of Artificial Intelligence. Shanghai
University Press.
6. Sun Weijun & Ni Shichao. (2020). Artificial Intelligence-Assisted Legislation
Under Major Public Health Emergencies. Shanghai Law Studies, 5

Publications and Papers in Other Countries

1. Cui Yadong. (2019, December). Artificial Intelligence and Judicial Moderniza-

tion. Springer-Verlag. Cui is the Secretary of the Leading Party Member’s Group
and President of the Shanghai Law Society.
2. Barış Soyer & Andrew Tettenborn. (2019, September). New Technologies, Arti-
ficial Intelligence and Shipping Law in the 21st Century. Professor Soyer is a
lecturer of Swansea, also the Director of the Institute of Shipping and Trade Law
at the School of Law, Swansea.
3. Fernando Iafrate. (2020, January). Artificial Intelligence and Big Data: The Birth
of a New Intelligence. Mr. Iafrate is the Head of Business Intelligence, Disney
Paris, France.

© Shanghai People’s Publishing House 2022 359

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2
360 Appendix C: List of Important Publications and Papers on Artificial Intelligence …

4. David Ekanem. (2020, March). Artificial Intelligence as a mechanism for crime

control in Nigeria: A Critical Appraisal.
5. Margarita Robles Carrillo. (2020, June). Artificial intelligence: From ethics to
law. She is a professor at the University of Granada, Italy
Appendix D
List of Important Conferences (Forums)
on Artificial Intelligence and Law
(2019.7—2020.6)

Conferences (Forums) in China

1. The 4th Annual Cyber Law Forum 30 of China & The 2nd Artificial Intelligence
and Law Education Forum. July 6, 2019, Chongqing.
2. Rule of Law Forum, World Artificial Intelligence Conference 2019. August 30,
2019, Shanghai.
3. Youth Rule of Law Forum, World Artificial Intelligence Conference 2019. August
30, 2019, Shanghai.
4. Legal Tech Conference 2019. September 8, 2019, Beijing.
5. The 3rd “Legal Big Data Research and Application” Seminar. September 21,
2019, Nanjing.
6. Global Law Forum (2019): International Conference on AI Policy and Law –
Algorithm Governance. October 18–19, 2019, Hangzhou.
7. Colloquium on Artificial Intelligence and the Future of Industry. October 19,
2019, Beijing.
8. The 2nd Colloquium on Big Data and Artificial Intelligence Technologies and
Related Legal Issues. November 9, 2019, Shanghai.

Conferences (Forums) in Other Countries

2019 OSCE Annual Police Experts Meeting: Artificial Intelligence and Law
Enforcement - An Ally or Adversary? September 23–24, 2019, Vienna, Austria.

© Shanghai People’s Publishing House 2022 361

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2
Appendix E
List of Typical AI Applications in the Judicial
Field (2019.7—2020.6)

This is a brief review of AI applications in the fields of legislation, law enforcement

and judiciary over the past year.

AI Application in Legislation

In Tianjin, an AI filing review platform was adopted in regulatory documents exam-

ination to ensure that legislation is in accordance with the law. In Shaanxi Province,
to improve legislative quality, new media and big data technology were used in the
process. In addition, Iran announced that it would take step in applying AI in legis-
lation. NATO also planned to apply AI to its decision-making process to enable the
efficient mobilization of data and personnel management. The White House has put
forward ten principles for AI regulation to avoid excessive government intervention
in AI development. The EU has issued a white paper on AI to guide the adjustment
of AI legislation.

AI Application in the Judiciary

In China, the Supreme People’s Court has inaugurated the Smart Court Laboratory,
and AI-assisted judicial practice, such as the AI judge system and intelligent prose-
cution system, has been established across China. During the COVID-19 pandemic,
courts in Guangdong Province implemented the “Public Security-Law Chain” in
evidence deposition and storage. The AI tools introduced by American courts were
also constantly optimized in the process of examination and approval practice. The
UK police used the HART system to determine whether a suspect should be detained.
In French, tax collectors successfully recovered 700 million Euros by AI algorithms.

© Shanghai People’s Publishing House 2022 363

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2
364 Appendix E: List of Typical AI Applications in the Judicial Field (2019.7—2020.6)

AI Application in Law Enforcement

In China, Shenzhen has adopted a new law enforcement mode of “AI + law enforce-
ment and case handling” and adopted AI for intelligent epidemic detection. Moreover,
the Ministry of Public Security will launch the Video Image Information Compre-
hensive Application Platform and Airport Smart Policing System in the near future.
Globally, police robots can be seen in dozens of countries, and the past year has
seen further applications of AI in policing. In the San Francisco Bay Area of the
USA, police used AI to select patrol locations. The British police planned to use
the National Data Analytics Solution system to arrest crimes. Calcutta police use
artificial intelligence technology (CMAPS) for crime investigation. In Dubai, the
police even used AI (Middle East Space Imaging SIME) to make predictions about
criminal activities.
Appendix F
List of Typical AI-Related Cases
(2019.7—2020.6)

1. Shenzhen Tencent Computer System Co., Ltd. v. Shanghai Yingxun Technology

Co., Ltd., A Dispute Over Copyright Ownership, Infringement, Commercial
Bribery and Unfair Competition.
The People’s Court of Nanshan District, Shenzhen, Guangdong Province, found that
the article in question was a written work protected by China’s copyright law and
belonged to a legal person’s work created under the auspices of Shenzhen Tencent
Computer Systems Co., Ltd.
2. Guo Bing v. Hangzhou Wildlife World. A Dispute over Personal Privacy.
In November 2019, the Fuyang District People’s Court accepted a lawsuit filed by
Guo Bing in October. The case was recognized as the “First Case of Facial Recog-
nition” in China; it was heard at the Fuyang District People’s Court and was put into
a second trial.
3. Chengdu Lunuo Information Technology Co., Ltd. v. Shanghai Hongzhong Exhi-
bition Service Co., Ltd. A Dispute over Infringement of the Right to Network
Dissemination of Information.
The case was heard online by Pidu District People’s Court and was recognized as
the first case that confirmed the effectiveness of electronic data through a judicial
blockchain platform in western China.
4. Guangzhou has implemented digital judicial authentication certificate forensic
opinion, which greatly improved the service and management efficiency of courts
during the COVID-19 pandemic.
The Guangzhou Municipal Bureau of Justice implemented the strategy of deploying
blockchain proposed by General Secretary Xi Jinping and fully utilized new tech-
nologies such as blockchain, facial recognition and electronic appraisal documents
to crack the challenges of judicial appraisal management.
5. The arbitration case between a private enterprise and a well-known state-owned
construction company
© Shanghai People’s Publishing House 2022 365
Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2
366 Appendix F: List of Typical AI-Related Cases (2019.7—2020.6)

Through facial recognition technology, parties concerned from four provinces and
five regions had a “no-touch” court hearing. The parties could continue to participate
in the arbitration process only if they had passed the facial recognition and submitted
identity documents, which ensured that the procedure was conducted in a safe and
confidential way.
Appendix G
Chronicle of Events of AI and Law

Since 2019, with the rapid development of AI technology as well as progress in

judicial reform and legal informatization, “AI + law” has become a growing trend
in innovation and entrepreneurship. Various applications and solutions of AI in the
rule of law have become more popular, gradually changing the operating mechanism
of the current legal system. Here is a list of big events in the field:
● The United States has released The National Artificial Intelligence Research and
Development Strategic Plan: 2019 Update.
● China has released Guidelines for National New Generation Artificial Intelligence
Open Innovation Platforms Construction Work.
● In China, Shandong Province has vigorously developed AI technology and made
progress in the construction of an “Internet + Healthcare” demonstration province.
● In China, the Ministry of Science and Technology has issued a Notice on the
Publication of the Guidelines for National New Generation Artificial Intelligence
Open Innovation Platforms Construction Work.
● Shanghai has issued Guidelines on Artificial Intelligence Security and Rule of
Law (2019).
● The World Artificial Intelligence Conference was held in Shanghai, China.
● The National Institute of Standards and Technology (NIST) formulated the
Guidance on Technical and Ethical Standards for Artificial Intelligence.
● The German Standards Institute (DIN) has published a German Standardization
Roadmap on Artificial Intelligence.
● The French Ministry for Economics and Finance has issued a Declaration on
Industrial Artificial Intelligence.
● In China, the “First Case of Facial Recognition” was heard in Hangzhou
● Russia has released a national AI strategy.
● China has released White Paper on Chinese Courts and Internet Judicial.
● In India, Tamil Nadu has announced policies on blockchain and AI.
● China released Opinions on Promoting the Development of “Internet + Social
Services”.

© Shanghai People’s Publishing House 2022 367

Y. Cui (ed.), Blue Book on AI and Rule of Law in the World (2020), Artificial Intelligence
and the Rule of Law, https://doi.org/10.1007/978-981-19-3586-2
368 Appendix G: Chronicle of Events of AI and Law

● In French, tax collectors successfully recovered 780 million euros by AI algo-

rithms “Fraud targeting and maximization of requests value” (CFVR in its French
acronym).
● China held the 3rd China AI Law and Ethics Forum.
● The Smart Court Lab of the Supreme People’s Court of China was put into use.
● The European Commission has revealed that it is considering a ban on the use of
facial recognition in public areas for up to five years.
● European Commission Launched the White Paper “On Artificial Intelligence - A
European Approach to Excellence and Trust” in Brussels.
● Technical University of Berlin, Germany, announced the establishment of a new
AI institute, “Berlin Institute for Learning Foundations and Data”.
● China’s National Bureau of Statistics accepted AI trainers as a new occupation.
● In China, Shanghai launched a reform on trial transcripts, and a “clerkless”
courtroom may appear in the near future.
● In Russia, an experimental legal regime was introduced in Moscow for a five-year
period.