Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Artificial Intelligence and Cybersecurity:

A Comprehensive Review of Recent Developments
Dr. Naveen Kumar C.G 1
Assistant Professor
Department of Computer Science Karnataka State Open University,
Mysuru, India

Abstract:- Due to the expansion of the internet of things  Objective

and connected devices, cyber security professionals in The research aimed to fulfil the following objectives:
the modern digital world face a significant number of
challenges. Experts require all available resources in  To study taking of artificial intelligence & cyber security
order to both respond to attacks and prevent further  How may AI change the state of cyber defence ?
security breaches. Due to factors such as heavy traffic,  To study tools for cyber security
an increase in the number of security attack vectors,
security vulnerabilities, and many others, the number of II. METHODOLOGY
connected workplaces renders it impossible for humans
to govern cyberspace without significant automation. Alongside the development of new technologies
However, designing software systems with comes an increase in the number and sophistication of
conventionally mounted algorithms (decision-making cybercrimes. Criminals operating online are using more
logic that is hardwired) can be difficult. These sophisticated methods to initiate assaults, which puts
algorithms are necessary to provide an effective defence contemporary security infrastructure in jeopardy. As a
against the growing number of network hazards. It is result, the cyber security sector is likewise undergoing
now common knowledge that adopting problem-solving change in order to keep up with the growing number of
strategies based on artificial intelligence is the key to security requirements posed by businesses. However, there
effectively addressing the vast majority of cyber is a possibility that these protective techniques used by
security issues currently being addressed. This article security specialists may fail at some time.
provides a concise analysis of cyber security
applications for computers. In addition, it investigates Artificial intelligence is being used by businesses in
the perspectives on how to enhance cyber security order to improve their detection methods for vulnerabilities
capabilities by providing applications that utilise and step up their overall game (AI). The use of artificial
artificial intelligence, in addition to the current intelligence in the field of cyber security is assisting
methods. businesses in protecting their defensive systems. In addition
to this, it helps them improve their analysis of cybercrimes.
Keywords:- Cyber Security; Artificial Intellegence; Expert
Systems. III. ARTIFICIAL INTELLIGENCE &
CYBER SECURITY
I. INTRODUCTION
Once upon a time, there was no connection between
The application of AI in security systems has the computer security and AI. Researchers in artificial
ability to lessen the rising and changing cyber security risk intelligence wanted to develop programmes that would
that is today faced by organisations all over the globe. As reduce the amount of labour required of humans, while
computing power, data collection capabilities, and storage those working in security wanted to prevent the leak of
capacity continue to advance, machine learning and artificial sensitive information. However, the two domains have
intelligence are being used in an increasing number of become increasingly intertwined as attacks have developed
industries and applications. This is a significant shift from to attempt to replicate the genuine performance not just at
previous years (AI). It is anticipated that this pattern will go the human user level but also at lower system levels. There
on. It is impossible for human people to comprehend this has been an increased convergence between the two fields.
large data gathering in real time because it contains too An excellent example of how AI might be used to bolster
much information. With the help of machine learning and safety is the "Completely Automated Public Turing test to
artificial intelligence, this mountain of data might be tell Computers and Humans Apart" (CAPTCHA). In order
reduced in a much shorter amount of time, which would be to complete the CAPTCHA, the user must enter the
of great assistance to the organisation in determining the characters that appear on a distorted picture. (2012) In
source of the security problem and recovering from it. It's addition to the typical login information, the user may be
possible that the employment of artificial intelligence in asked to input a secret code made up of letters and digits.
security teams will radically change the game. (2012) Enhanced ACR software may be seen as a major

IJISRT23MAY1138 www.ijisrt.com 1928

 Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
breakthrough in AI, inspiring a transition to more intricate and digesting the high-level raw data, much as our brain
pattern recognition. For this reason, the commercial learns on its own from the raw data utilizing the inputs
security industry is indirectly promoting the development from our sense organs. When this kind of deep learning,
of artificial intelligence (AI) by working to safeguard which has been taught, is used to computer security, the
activities like online ticket purchases. Our solutions rely system is able to determine, without the need for human
heavily on AI because of its ability to rapidly identify and interaction, whether a file is harmful or lawful. In
assess new exploits and holes, reducing the severity of comparison to traditional methods of machine learning, our
future assaults. approach yields much better results while searching for
malware. The rapid speed at which neural networks may be
The ability to identify and react to previously implemented in hardware or graphics processors is the key
undiscovered threats is a major draw for the use of artificial to their success in the field of cyber security. The exact
intelligence (AI) techniques in intrusion detection. For identification of emerging malware threats and the filling in
example, when it comes to understanding particularly novel of important gaps that leave businesses vulnerable to
types of cyber-attacks, artificial intelligence systems that assault are both capabilities that may be enabled by neural
are programmed to learn and adapt and can detect even the nets. (Gostev, 2012)
smallest of changes in their surroundings can respond much
more quickly - and on the basis of vast troves of data - than  Intelligent Agents
humans can. (Dunn Cavelty, 2018) An intelligent agent (IA) is a self-sufficient organism
that monitors its surroundings with the use of sensors,
 Expert Systems responds to its surroundings by employing actuators (thus
One kind of artificial intelligence, or AI, is a "expert the name "agent"), and focuses its actions with the ultimate
system," which is a computer programme designed to aim of accomplishing certain objectives. To accomplish
mimic human experts' level of expertise and decision- their objectives, intelligent agents may acquire new
making skills. Knowledge-based systems like this one are information or make use of existing information. They
composed of two components: a knowledge base and an might be quite simple or extremely complex; one example
inference engine. The assertions and examples that make up of an intelligent agent is a reflex mechanism like a
the knowledge base are a portrayal of the world as we know thermostat. They exhibit the following behaviors:
it. An automated reasoning system, the inference engine responsiveness, pro-activeness, and a comprehension of the
examines the current state of the knowledge base, agent communication language. They are able to adjust to
implements the rules that are relevant to that state, and then real-time situations, pick up new skills rapidly via
asserts fresh information into the database. The Cyber engagement with their surroundings, and have the capacity
Security Artificial Intelligence Expert System, often known to store and retrieve examples from memory. In order to
as CSIA, is comprised of the following essential elements defend against Distributed Denial of Service (DDoS)
in its knowledge base and Inference Engine: In order to assaults, an intelligent agent is being created. The creation
protect against cyber attacks, the Security expert system of a "cyber police" force that is equipped with mobile
goes through a series of processes. It compares the process intelligent agents ought to be feasible in the event that there
to the knowledge base to determine whether or not it is a is a conflict involving legal or economic matters. Because
well-known process that should be ignored; if not, the of this, we need to put in place the infrastructure that will
system should end the process. If there is no such process support the quality of the intelligent agents and their
in the knowledge base, the expert system will use inference communication with one another. The use of multi-agent
engine methods, also known as rule sets, to determine the technologies, such as a hybrid multi-agent and Agent-based
current state of the machine. The condition of the machine distributed intrusion detection, will provide the cyber police
has been broken down into three distinct categories: safe, with a more comprehensive and effective operational
moderate, and severe. According to the current condition of appearance. (Raiu, 2012)
the machine, the notification is sent to the administrator or
user, and the knowledge base has been updated with the
inference. (Bradbury, 2021)

 Neural Nets
Deep learning, which sometimes refers to neural
networks, is an advanced subfield of artificial intelligence.
The processes of the human brain served as an inspiration
for it. Our neurons, which for the most part have no
specialized function and are independent of certain
domains, have the ability to learn any kind of information.
Frank Rosenblatt invented the artificial neuron known as
the Perceptron in 1957, which paved the path for the
development of neural networks. Through interaction with
other perceptron’s, these ones are able to learn and find
solutions to challenging problems. They learn to recognize
the thing on which they are taught on their own by learning Fig 1 Cyber Attacks

IJISRT23MAY1138 www.ijisrt.com 1929

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
IV. HOW MAY AI CHANGE THE STATE OF
CYBER DEFENCE

Companies are placing an unprecedented amount of

emphasis on the need of cyber security right now. This is
due to the fact that sophisticated cyber-attacks have
resulted in data breaches that have cost businesses millions
of dollars. The first step is to design a multi-layered
security system that will be responsible for protecting the
network infrastructure. Installing a firewall that will block
unwanted network traffic is the first thing that has to be
done.

After that, antivirus software is applied to the Fig 2 AI in Cyber Security

infrastructure in order to remove any dangerous files and
viruses that may be there. Data is backed up on a consistent  Tools for Cybersecurity
basis as part of their contingency strategy for handling These days, businesses are also using machine
emergencies. (Bockus, 2015) learning, a kind of AI, to bolster the safety of their internal
networks. Its primary use is assisting security professionals
Artificial intelligence has had an effect on network in identifying malicious assaults, but it also has the
security by assisting experts in spotting anomalies in the following uses:
network via the study of trends and the analysis of user
behaviors. AI now makes it possible for security experts to
analyses network data in order to identify weaknesses and
thwart malicious attempts. The following are some of the
ways that AI will assist to improve the conventional
security approach:

 Monitoring and responding to security incidents will be

done with the help of cutting-edge security solutions
driven by artificial intelligence.
 The most up-to-date firewalls will come equipped with
machine learning technology that can instantly identify
any deviation from the normal flow of network traffic
and block it if it is deemed to be harmful.
 Professionals in the field of information security are
able to determine the origin of a cyberattack by using
the natural language processing function of AI. The use
of natural language processing is also beneficial for
doing vulnerability assessments. Fig 3 Framework for Cyber Security Tools
 By conducting data searches throughout the internet
and making use of predictive analysis, dangerous  Protection for mobile endpoints
threats may be found ahead of time. Mobile endpoint security uses machine learning since
 Enhanced safety via the use of authentication and smartphones, tablets, and laptops are all vulnerable to
controlled access cyberattacks. Recently, Wandera released their MI: RIAM
threat detection engine, which is driven by machine
Biometric sign-in systems are an additional significant learning. This engine has found many samples of mobile-
breakthrough brought about by artificial intelligence in the focused SLocker Ransomware that has been repackaged.
field of cyber security. The usage of fingerprints, retina
scans, and palm prints are all components of these very safe  There are no known exploits with zero-day potential.
login methods. (Alsmadi, 2019) Along with these To put it simply, a zero-day vulnerability is a kind of
fingerprints or other biometric data, you are also able to log security risk for which no known workaround exists. A
in securely by using a password. This technology is utilized zero-day vulnerability is one that has been discovered by
in businesses to allow workers to log in, and it is even security researchers but has not yet been exploited by a
integrated into certain smartphone operating systems. malicious actor. Unprotected Internet of Things gadgets
might occasionally harbor such dangers. (CRUZ LOBATO)

Algorithms trained with machine learning may spot

zero-day threats by looking for unusual behavior in
network data. Using machine learning, vulnerabilities are
patched out and exploits are avoided.

IJISRT23MAY1138 www.ijisrt.com 1930

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 Optimizing Analysis by Humans [5]. CRUZ LOBATO, L. U. Í. S. A. (n.d.). Unraveling
Vulnerability assessment, threat detection, network the cyber security market: The struggles among
analysis, and endpoint security are all areas where human cyber security companies and the production of
analysis may be supplemented by machine learning. Cyber (in)security. https://doi.org/10.17771/
Machine learning algorithms may identify potentially pucrio.acad.27784
harmful information in a network and forward the results to [6]. Cyber security 2019. (2019). 2019 International
a human security expert. Thus, there is potential for much Conference on Cyber Security and Protection of
improved warning detection rates. Digital Services (Cyber Security). https://doi.org/
10.1109 /cybersecpods.2019.8885065
 Eliminating Human Error from Security Procedures [7]. Cyber security 2020 cover page. (2020). 2020
With the use of machine learning, tedious security International Conference on Cyber Security and
duties may be automated away. Working in this manner Protection of Digital Services (Cyber Security).
allows experts to better priorities tasks. Machine learning https://doi.org/10.1109/cybersecurity49315.2020.91
can automate a wide variety of tasks, including monitoring 38881
network traffic, blocking threats like ransomware, cleaning [8]. Cyber security evolution. (2012). Cyber Security
up infections, and analyzing network data. (Goyal & Policy Guidebook, 15–38. https://doi.org/10.1002
Rajput, 2020) /9781118241530.ch2
[9]. Cyber Security Objectives. (2012). Cyber Security
Machine learning may also be used to effectively Policy Guidebook, 39–67. https://doi.org/ 10.1002
allocate human security resources. /9781118241530.ch3
[10]. Dunn Cavelty, M. (2018). 27. cyber-security.
V. CONCLUSION Contemporary Security Studies, 410–426.
https://doi.org/10.1093/hepl/9780198804109.003.00
In light of the recent rise in malicious software and 27
hacking attempts, the implementation of an intelligent [11]. Gostev, A. (2012). Cyber-threat evolution: The Year
security system has become essential. AI approaches are Ahead. Computer Fraud & Security, 2012(3),
more adaptable and resilient when compared to 9–12. https://doi.org/10.1016/s1361-3723(12)70052-
contemporary cyber security solutions; as a result, they 0
increase security execution and better guard systems from [12]. Goyal, D., & Rajput, R. S. (2020). Cloud
an expanding variety of sophisticated cyber threats. Despite computing and security. The Evolution of Business
the profound shift that AI has brought about in the field of in the Cyber Age, 293–319. https://doi.org
cyber security, associated frameworks are not yet prepared /10.1201/9780429276484-12
to totally transform and, as a result, adapt to changes in [13]. Raiu, C. (2012). Cyber-threat evolution: The past
their state. Although there are numerous advantages to year. Computer Fraud & Security, 2012(3), 5–
using AI methods for cyber security, it is important to keep 8. https://doi.org/10.1016/s1361-3723(12)70051-9
in mind that AI is not the sole solution to security [14]. Security and Trust in Cyber Space. (n.d.). Cyber
problems. The intelligent security system will be rendered Law and Cyber Security in Developing and
ineffective when it is attacked by a human adversary with Emerging Economies. https://doi.org/10.4337
the intention of circumventing it. This doesn't imply we /9781849803380.00005
shouldn't use AI approaches; nevertheless, we should be
aware of the constraints they have. AI requires ongoing
engagement and training from human beings. This
combined strategy has shown success on numerous
occasions, and it collaborates with security experts in an
effective manner.

REFERENCES

[1]. Alsmadi, I. (2019). Cyber Security Management.

The NICE Cyber Security Framework, 243–251.
https://doi.org/10.1007/978-3-030-02360-7_10
[2]. Behavioural science in cyber security. (n.d.). Cyber
Security: Law and Guidance.
https://doi.org/10.5040/9781526505897.chapter-027
[3]. Bockus, N. F. (2015). Cyber in space: 2035.
Advances in Information Security, 39–57
[4]. Bradbury, R. (2021). Educating for cyber (security).
The Oxford Handbook of Cyber Security, 394–408.
https://doi.org/10.1093/oxfordhb/9780198800682.01
3.24

IJISRT23MAY1138 www.ijisrt.com 1931