ARTIFICIAL INTELLIGENCE AND SECURITY. SECURITY 4.

0
Prof. Radulov, N., PhD
Department National and International Security, New Bulgarian University, Sofia, Bulgaria
[email protected]

Abstract: The Artificial Intelligence (AI) has giant possibilities to optimize the fight against crime and strengthen national security. In the
conditions of unimaginable accumulation of information and the need for rapid decision-making, only the use of AI can lead to success.
Intelligence, counterintelligence, forensic science, counteracting organized crime, rapid processing of available information, drafting of varied
decisions, creating plans and multivariate scenarios, performing various analyzes is a time-consuming process. Only its use can significantly
shorten this time and thus dramatically increase the possibilities for detection, prevention and curbing crimes.
Keywords: Artificial Intelligence; Intelligence; counterintelligence; security; forensic science; investigation; crime.

1. Introduction interest of man. The basic idea is that the purpose of the machine must
be concluded in a maximum outreach of beneficial for people targets,
Artificial Intelligence (AI) can turn into reality the scenarios of but it does not have to know in advance what they are. In addition to
today's fantastic movies: intellectual operating systems and digital being useful for the people, the solution must be guaranteed in its
assistants. It may even be possible for robots to perform basic police security and safety. New technologies, such as quantum computing,
functions. In practice today, if the police have high-quality computer may change the AI's approach to getting information about different
systems with the scope of an AI, much of their routine office work issues, and allow it to learn by receiving feedback, and perhaps even
can be done by them, resulting in new reservations being made to imitate human cognitive contact functions with the world. If this
increase police presence in urban areas. One of the main resources happens, AI will bring economic benefits, working without inherent
today - the human - in police activity is increasingly burdened with human errors due to uniformity and cumulative fatigue. Giant
administrative and cabinet activities, which leads to fewer police information systems with security-related information stored in them
officers on the street. AI is now in a position to ensure the monitoring can serve as a basis for comparing, improving and intensifying the
of video-data flows and data collected from a large number of sensors deep learning process in a specific area such as national, international
and can warn security services of suspicious activity. It is not far the and civil security. There are many routine procedures in security
complete integration of nowadays CCTV networks, their synthesis processes that AI can handle with ease, but it is important to have a
with modern image search programs, the parallel inspection of all proactive and creative action with preemptive nature. Such action,
available databases of people with contact with compromised based on a series of AI-proposed management solutions due to
persons, their ranking according to predefined indicators developed electronic forecasting, observed and corrected by talented managers
by experts. Today the police is already using robots to conduct search and analysts, will lead to a major breakthrough in countering modern
and rescue operations 1, to dispose explosive devices in terrorist security threats in its broad and narrow sense. Early identification of
activity 2 and even to destroy armed criminals 3. threats, accurate forecasts in a different timeframe, provided
2. Integration of AI into our familiar world sufficiently and timely, backed up with detailed multi-factor
scenarios of management decisions, can dramatically optimize the
Today, Orbital Insight is practicing machine self-training for low fight against crime worldwide.
resolution photos from Landsat (US) and Sentinel (in the EU)
satellites. It leads to easier identification of military sites, terrorist Security issues are also the issues of the safety and vulnerability
camps, solving problems of domestic and international security, of AI. Although the specialized AI offers vast opportunities to
counteracting organized crime, for example, monitoring and tracking society, it can also be deceived, penetrated, or misled. Breaking the
channels for human trafficking and smuggling of goods. security systems is doubly devastating because it affects not only
themselves but also the quality of their product – national, public and
Practice shows that the largest amount of funds are invested in civil security. We need to be sure that the decisions taken by the
programs in the national security sphere of the individual countries – machine are not the result of external interference and can not be
since September 11, 2001, over $ 500 billion were given for NSA's changed by cyber attacks.
electronic intelligence systems, and most high-tech products
originated from military projects - for example, ARPA-Internet. 4 This 4. Features and capabilities of AI
creates an additional and serious threat to predominantly In order to be aware of the security-related AI capabilities, we
militarization of AI by individual countries and individuals with the need to consider the following features.
necessary assets. It is quite possible that some organized crime circles
allocate funds and gain access to AI and specialized military and spy- The AI changes over time. Today, artificial intelligence is
oriented robots 5. commonly referred to as machine learning opportunity -
programming approaches using different algorithms and methods
3. AI self-learning (Deep Learning) such as linear regression, decision trees, Bayesian networks 6,
We must proceed from the fact that adequate quality system can evolutionary algorithms, and artificial neural networks. Our
solve every task assigned to it, so it is necessary to formulate such a understanding of what is AI varies with the passing of each milestone
task that the solution found by the machine is unquestionable in the in this field. Adaptability, flexibility, predictability and proactivity in
terms of minimum time resources, the speed of decision-making and

1
https://www.intechopen.com/books/search-and-rescue-robotics-from- turnover controlled by criminal groups. Generally speaking, criminal groups
theory-to-practice/introduction-to-the-use-of-robotic-tools-for-search-and- control not less than a third, but rather about 50% of the global turnover of all
rescue kinds of goods and services, assets and finances. – Goodman, Marc. Future
2 Crimes. Inside the Digital Underground and the Battle for Our Connected
https://www.hsdl.org/?abstract&did=484061 World, Penguin Random House, 2016.
3
https://www.uclalawreview.org/policing-police-robots/ 6
Bayesian networks, also known as "belief networks" or "causal networks",
4
Investments in the robot industry in 2019, according to experts, will exceed are graphical models for representing multidimensional probability
$ 135 billion, almost twice as high as in 2015 (n. a.) distributions. – http://www.cs.cmu.edu/afs/cs.cmu.edu/project/learn-
5
It can now be estimated that transnational crime accounts for 15-20% of the 43/lib/photoz/.g/web/glossary/bayesnet.html
world gross product as purely criminal turnover and not less than 25% as legal
scenarios realizing them should be the priorities of the AI used in the artificial intelligence analysis methods. Identifying the signs of crime
security sphere. is the first step for further investigation. Using this approach, crime
information can be automatically collected and entered into law
AIs, robots and people work better when it comes together, there enforcement databases 8.
is a factual synergy. In practice it has already shown that people who
carry out complex actions or play games are more effective and win AI, training, deep learning and fighting crime
when interacting with high-tech applications than when analogous
activity is done only by people or just by computers. Increased Machine Learning - computer algorithms that allow AI to learn to
creativity of man helps to train the AI so that it strives to create work with a large amount of data are widely used by technology
original solutions using its capabilities quickly process huge data sets, giants such as Google, Apple and Netflix. The same technology that
classifying and arranging a giant number of operations, operative allows programs to recommend films or to set the order in the Google
decisions and scenarios that allow special services to implement them search results list is now successfully used to combat crimes of
successfully. In the field of security, it is clear that this, who apply varying severity.
high technology, uses more, better-processed and verified Detecting crimes
information, has a wider range of standard management to which he
can rely. He receives more detailed scenarios for strategic and Because AI is able to analyze a huge amount of data, it can be
operational solutions, and his control capabilities are enhanced. This used to both investigate and prevent crime. An example of this is the
presupposes a natural basis for creative thinking and management advertising of sex services on the Internet. Sex workers are thought
therefore intelligence service gets a great resource for anticipating to be relatively anonymous and advertise their services almost under
impact and success. the nose of law enforcement authority. But every message on the web
leaves a digital trace. Cyber experts can use algorithms for digital
There are AI systems that help leverage data obtained from open tracking data on the internet to train AI to see if women volunteer or
sources, but their use for machine training needs to be arranged and coerced advertise their services, focusing on seemingly insignificant
protected accordingly. Data generated by specific forces and means, details. By processing the input data, algorithms are programmed to
as well as by processing and monitoring of publicly available sources, analyze such details as a boom of small theft arrests in the area where
are equally actively used in the security field. Allen Dulles claimed sex services are advertised. AI may detect an increase in thefts of the
that intelligence works with over 90% acquired from open sources least important things, such as toothpaste or soap. The idea of this
data 7. Therefore, the AI in the security can use all the possibilities methodology is that if someone is a victim of human trafficking and
for collecting, sorting and managing the data existing in the is brought from another region, he may be deprived of access to basic
information space, which optimizes the performance of a quality, subjects for a normal life. Furthermore, AI training is applicable to
timely and effective activity in providing national and civil security. data related to hotel rooms paid in cash in areas where sexual services
Even the best intelligent systems can make mistakes and are provided and advertised during major events such as concerts or
deviations. In the field of security, the data obtained should be further sporting events. This algorithm applies to other details related to
verified in order to eliminate the possibility of both incompetence and trafficking in human beings. For example, analyzing the above-
subsequent crisis and possible deliberate misinformation resulting mentioned messages, law enforcement officers use the training
from an enemy's operation. opportunities of AI to establish links between ads and their authors,
paying attention to such nuances as similar styles of writing and
The security impact of AI will depend on the approach and how settings used in more than one ad , on websites in missing persons'
it is used. The practical application for solving real problems will be databases. A learning algorithm is also used to track payment
decisive. Increasing the capabilities and influence of intelligent transactions to provide intimate services to identify the links of those
systems and robots will increase the importance of creative decisions paying for advertising with larger organizations involved in
taken by executives in the sector about location and time of use. This trafficking. Machine learning technology for AI can be applied not
will require a radical change in the security manager's profile only to the detection of crimes related to human trafficking and forced
regarding his management culture – the set of knowledge, habits and prostitution but also for practically any other types of crime.
skills, as well as a change in the subjects he will study and practice –
cyber-management, cyber-planning, cyber- forecasting, cyber- Forced labor
scenarios, etc. By training and self-learning AI, law enforcement can determine
whether some business works together with companies using forced
5. Artificial Intelligence and Counteracting Crime
labor. In the area of supply chain logistics, the use of forced labor is
Collecting information not unusual. AI can help investigate allegations of labor violations or
complaints against companies using forced labor. Analysis of data
Collecting data on crime situation is a way to derive the necessary through AI may reveal links that would otherwise be difficult to
knowledge from large data sets. In other words, this is an approach to identify.
discover hidden relationships between organizations and individuals
committing crimes by using artificial intelligence methods. The wide Cybercrime
range of data mining applications has become a significant area of
research nowadays, as part of these applications are related to AI can help companies prevent cyber-crime that can cause
detection, prevention and curbing crimes. We must not forget the financial damage and harm their reputation. It can be trained to
postulate that information and its analysis lie at the heart of the recognize keywords or topics related to harmful content, thus
successful fight against crime, as well as the management of this stopping the potential cyber attack.
counteraction. Burglary
Criminology Law enforcement can use training to detect crimes such as
Criminology is one of the most important areas of application of burglary. For example, data on a large volume of theft (date and time
intelligent data analysis. It is a process whose purpose is to identify when offenses were committed, crime objects and methods, tools to
the criminal characteristics. In fact, crime analysis involves commit) may show similarities with other comparable crimes that
investigating and detecting crimes and linking them to criminals. The have not been resolved. After gathering information about the crime
large numbers of crime data and the complexity of links between them (investigation of the crime scene, interviewing witnesses,
have turned criminology into a suitable area for the application of investigating crime objects, etc.), they can be used to train AIs to

7 8
The Craft of Intelligence, 1963, Lyons Press; Reprint edition, May 1, 2016 https://www.forbes.com/sites/bernardmarr/2017/09/19/how-robots-iot-and-
artificial-intelligence-are-transforming-the-police/#58325e805d61
analyze this data, find links with others crimes. AI can easily detect it 9. www.uclalawreview.org/policing-police-robots/
when formalizing a large enough volume of crime data, or identify
suspects and identify gaps in the investigation, and at the same time 10. https://watermark.silverchair.com/labmed39
relieve police officers of unnecessary work to be able to focus on
other important aspects of the investigation.
29.03.2019
DNA analysis
Sofia
Modern forensic expertise of DNA is critical to the investigation
of crimes, but the interpretation of the DNA code can be quite
complicated. Specialized AI training techniques can be used to
simplify the interpretation of DNA, especially when it comes to DNA
samples of several individuals. There is a huge amount of data that is
not currently read solely because our limited capabilities, but the
computer algorithms of AI can easily do this 9.
10
Face recognition
A common method of combating crime using AI is the Face
Detection Technology. It is often used at airports to include mapping
of human images into law enforcement database files, which allows
to identity the perpetrator.
Prevention of crimes of a terrorist nature
Scanning social networks to find people who can be radicalized
is another activity that AI can effectively perform after appropriate
training. Nowadays, some law enforcement agencies are already
using social networking monitoring and analysis to prevent attempts
to recruit new members of terrorist organizations such as ISIS and
others like them 11. One such monitoring tool is called iAWACS or
the AWACS internet system. This name is similar to the abbreviation
used by the US military to describe its intelligence stations. The
purpose of iAWACS is to prevent negative events by monitoring the
activity on the Internet, identifying and locating potential criminal
scenarios.
6. Conclusion
The abovementioned abilities of artificial intelligence in the
security sphere impress with the breadth of application and the
possible positive effect. Without expanding the scope of the solutions
discussed, which can be implemented with the help of the AI, we can
certainly argue that it and Security 4.0 will completely change the
possibilities for enhancing the quality and intensity of the product
“civil and national security” and this change is close and forthcoming.
Bibliography
1. Dulles, A., The Craft of Intelligence, 1963, Lyons Press;
Reprint edition, May 1, 2016
2. Goodman, Marc. Future Crimes. Inside the Digital
Underground and the Battle for Our Connected World, Penguin
Random House, 2016
3. www.cs.cmu.edu/afs/cs.cmu.edu/project/learn-
43/lib/photoz/.g/web/glossary/bayesnet.html
4. www.forbes.com/sites/bernardmarr/2017/09/19/how-robots-
iot-and-artificial-intelligence-are-transforming-the-
police/#58325e805d61
5. www.hsdl.org/?abstract&did=484061
6. www.intechopen.com/books/search-and-rescue-robotics-
from-theory-to-practice/introduction-to-the-use-of-robotic-tools-for-
search-and-rescue
7. www.roboticsbusinessreview.com/ai/ai-advances-facial-
recognition-robots/
8. www.smartdatacollective.com/future-social-media-depend-
artificial-intelligence/

9 11
https://watermark.silverchair.com/labmed39 https://www.smartdatacollective.com/future-social-media-depend-
10
https://www.roboticsbusinessreview.com/ai/ai-advances-facial- artificial-intelligence/
recognition-robots/