ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
175
MAIN ISSUES IN THE SPHERE OF INTERNATIONAL LEGAL
REGULATION OF ARTIFICIAL INTELLIGENCE
Khamdamova Firuza Urazalievna
Doctoral candidate of the National Center
of the Republic of Uzbekistan for Human Rights
Doctor of Philosophy in Law (PhD)
Uzbekistan, Tashkent
E-mail: fkhamdamovaphd@mail.ru
https://doi.org/10.5281/zenodo.13853866
Abstract
: The article is devoted to an overview of key issues related to the
legal regulation of artificial intelligence. The author notes that there are no
unified positions regarding the legal regulation of AI. One group of scientists
considers the legal regulation of AI necessary, while others warn that excessive
regulation may hinder the development of AI. The article highlights the
following key issues: the content of the concept of "artificial intelligence" (AI),
the legal personality of AI, i.e. the issue of recognizing AI as a subject of law, the
liability of AI in the event of harm. The author notes that at the moment there
are many definitions of AI, the issue of recognizing AI as a subject has not been
resolved unambiguously, however, the so-called theory of fictitious legal
personality of artificial intelligence prevails, which assumes giving the legal
status of AI certain features inherent in the status of legal entities. The article
provides examples of legal regulations in the EU, as well as examples of cases
from judicial practice on the recognition of AI as a subject of law.
Keywords:
artificial intelligence, electronic person, legal personality of AI,
fictitious legal personality of AI, liability of AI.
At the moment, many different acts devoted to AI ethics have been
adopted around the world within the framework of international organizations
and in individual countries. In order to ensure clear legal regulation of activities
in the field of AI, it is initially necessary to define the legal content of the term
"artificial intelligence" itself, as well as to identify its basic features. Despite the
intensive development of AI and the dynamic development of scientific debate in
this area, there are still no unified approaches to defining AI. One reason for the
lack of a unified definition of the concept of "artificial intelligence" is the
interdisciplinary nature of approaches to understanding this issue, since the
implementation of scientific research in the field of artificial intelligence
requires the unification of knowledge from various scientific fields. [5]. In other
words, the definition of artificial intelligence depends on the discipline in which
the phenomenon is considered.
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
176
One of the definitions was proposed by the High-Level Expert Group on
Artificial Intelligence (AI HLEG) of the European Commission within the
European Union in a report presented on April 8, 2019 [8]. In this document, the
Group distinguishes between the concepts of artificial intelligence systems as
such and artificial intelligence as a scientific discipline. Artificial intelligence
systems are defined in the report as "systems that exhibit intelligent behaviour
by analysing their environment and taking actions — with a certain degree of
autonomy — to achieve specific goals." There are many other options for
defining the concept of "artificial intelligence", however, there is no generally
accepted definition enshrined in an international legal document. At the same
time, it is possible to note the similarity of researchers' opinions regarding the
basic characteristics of artificial intelligence, in particular, that systems based on
artificial intelligence are systems that make critical decisions without human
participation.
The European Parliament resolution "Civil Law Rules on Robotics" of
17 February 2017 notes the need to develop common concepts for such terms as
"cyber-physical systems", "intelligent autonomous robots" taking into account
the following features:
1) the ability to be autonomous and exchange data; 2) the ability to learn
from experience;
3) the presence of minimal physical support;
4) the ability to adapt to the external environment;
5) the absence of life [2, C.26].
It is important to emphasize that to date, no official document contains a
normative definition of the concept of "artificial intelligence", although the term
itself is actively used in many countries, along with such terms as "robots". Thus,
modern science has not developed a uniform understanding of artificial
intelligence. Today we can talk about many definitions of artificial intelligence,
since the areas of its application are very wide.
The above raises the question of determining the content of the concept of
"artificial intelligence" and its recording in legal texts. The lack of a clear legal
definition of artificial intelligence hinders the resolution of other legal issues, in
particular such issues as the legal personality of artificial intelligence, liability
for harm caused by the use of artificial intelligence. Thus, the military use of
artificial intelligence violates the most basic natural and inalienable human right
- the right to life. Currently, artificial intelligence is developing in almost all areas
of activity of the armed forces of many states.
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
177
Another example - the implantation of cyber-physical systems into the
human div is already becoming a reality today. The European Parliament, in its
report of 27.01.2017 with recommendations to the European Commission on
the civil-law regulation of robotics, draws attention to the risk associated with
the possibility of hacking or disabling operational programming systems
embedded in the human div, or erasing their memory, which can endanger
human health, and in special cases even their life..[4, C.70].
AI poses a threat to human rights. Another example is the use of artificial
intelligence units in public administration, which initially undermines the idea of
governing people by people based on democratic procedures and principles. In
addition, artificial intelligence units can be successfully used in the judicial
process, including for the analysis of judicial practice and the substantiation of
judicial decisions. This can simultaneously lead to a violation of the right of
citizens to a fair trial. Acceleration of the consideration of court cases in such a
case does not always lead to the humanization of judicial activity and
compliance with the principle of justice. [4, C.72]. In addition, the use of artificial
intelligence technologies allows for open interference in private life, nullifying
this recently won right of people. [4, C.75].
In this regard, a set of issues regarding the legal regulation of AI has been
formed to date.
The first issue is the debate about the possibility of recognizing/not
recognizing artificial intelligence as a person. A number of researchers believe
that the legal status of an electronic system with elements of artificial
intelligence and a perfect autonomous unit cannot be the same. The latter, in
their opinion, can be recognized without unnecessary hesitation as a full-fledged
cyber subject of society, but with the proviso that such a status will have
different options in terms of the range of rights and responsibilities, since it is
not possible to put an ATM, a smart home system and a combat robot in the
same row. From this, they conclude that systems with artificial intelligence must
have a certain legal status, which will depend on the functionality and other
features of a particular system. [1].
When discussing the issue of the possibility of recognizing artificial
intelligence as a subject of law, two options are put forward. In the first case,
artificial intelligence is understood only as a technical means with the legal
regime of a thing. In the second case, it is recognized as an electronic person by
analogy with a legal entity through the use of a legal fiction. Both options are not
fully adequate.
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
178
The qualification of artificial intelligence as an object of law does not take
into account the ability to think and make independent decisions. In the second
case, artificial intelligence is equated with a person similar to a human. The
model of legal regulation depends on the solution to this issue, starting with the
possibility of entering into legal relations and up to imposing legal liability on
such intelligence.
At the same time, it is obvious that for artificial intelligence to acquire the
status of a subject of law, it must have such a quality as will. Artificial
intelligence does not have volitional ability. Therefore, granting legal personality
to artificial intelligence will in any case be a fiction
[2, C.26].
The most stable
and progressively developed in scientific works is the theory of fictitious legal
capacity of artificial intelligence, which assumes that the legal status of AI will be
given certain features inherent in the status of legal entities.
As a result of this controversy, the use of the term "electronic person" was
proposed, and the idea of endowing it with legal capacity is increasingly heard in
the European Union. The draft report on the status of the "electronic person"
developed by the European Parliament Committee on Legal Affairs indicates
such signs of "intelligence" of robots as the ability to analyze data; the ability to
adapt their behavior; the presence of physical support; autonomy acquired
through sensors and contact with the environment, as well as the ability to self-
learn. The document notes that the most highly developed and high-tech robots
should acquire the status of an electronic person with its inherent legal capacity,
and this status should always be used when robots independently make
volitional decisions or otherwise interact with third parties. Such approaches
directly indicate the desire of a part of society to consider the "electronic
person" as a real actor, since it is assigned this or that duty, which it can fulfill or
not. In this regard, the question arises: who will be responsible for failure to
fulfill this duty? And there is no unambiguous position here. At the same time,
there are other points of view on this issue. Thus, the American professor L.
Soulum has already formulated theses proving the legal groundlessness of
recognizing that artificial intelligence has the status of a person. He believes that
electronic systems, and even systems with full artificial intelligence, cannot be
considered as entities similar or identical to people. And as an example, he cites
the 14th Amendment to the US Constitution, according to which all persons born
or naturalized in the United States, being subjects of the jurisdiction of such, are
citizens of the United States. Thus, only people can be born, therefore, artificial
intelligence cannot have the rights of citizens [6, C.96].
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
179
The second question is a question that follows from the question of
recognizing AI as a subject - the problem of liability for harm caused by such
intelligence. In legal literature, various models of imposing tort liability are
discussed:
• on the owner of the rights to a device equipped with artificial
intelligence;
• on the software developer;
• on the operator servicing artificial intelligence [2, C.26].
Currently, one of the pressing issues, which has both theoretical and
practical significance, is the possibility of recognizing robotics as a subject of
law. Obviously, today it is difficult to underestimate scientific and technological
progress, which marked the emergence of artificial intelligence (AI) and other
technologies. In this regard, the possibility of recognizing a robot with artificial
intelligence as a subject of law, legislative consolidation of its legal status, as well
as the consequences of the development of robotics are discussed. The existence
of the need to recognize artificial intelligence as a subject of law is a debatable
issue. The possibility of assigning a robot the status of a legal entity is a rather
controversial issue.
Most lawyers believe that the robot should be given the status of an
"electronic person". Since it is an object of increased danger, it must have an
operator - a "guardian" (the owner of this robot), who will be responsible for the
actions of the work, but the robot will have a number of rights.
Without an operator, the robot will not be able to perform a number of
certain significant actions. Artificial intelligence operates on the basis of self-
learning, i.e. the operator, in turn, will be able to contribute to the behavior of
the robot. Not every robot will fit these rules, for example, an exception may be a
robot vacuum cleaner, which simply performs the function of cleaning floors.
Each robot will have to be registered under a certain identification number.
Failure to register a robot may entail both administrative and criminal
punishment. Criminal punishment may be applied only if this robot contradicts
the principles of humanism, morality, etc. Registration cannot be made if the
robot does not have certain functions, namely: an emergency shutdown button,
an action cancellation mode, etc. In this case, the operator will have to pass a
special commission for access to ownership of such a robot in order to somehow
exclude the fact that the robot will be used to create conditions for committing
actions prohibited by law. In order to avoid the emergence of any autonomous
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
180
combat system that can kill on its own initiative, AI must necessarily be guided
by certain rules that cannot be replaced by interference in the system.
Case law on AI issues is gradually developing, in particular cases on
recognizing AI as legal entities. For example, the DАBUS case in the USA. Dr.
Steven Thaler created a device called DABUS, which consists of neural networks
and was used to invent an emergency warning light, a food container that
improves grip and heat transfer, and much more. Thaler believes that artificial
intelligences should be patent holders and has launched numerous lawsuits in at
least 15 countries to prove his case. The US Court of Appeal once again
confirmed previous decisions that artificial intelligence cannot have invention
patents. So far, all cases in the UK, US, and New Zealand have been lost. In
Australia, Ryan Abbott, a professor at the University of Surrey, filed patent
applications on behalf of Thaler. In August 2021, the Federal Court of Australia
ruled that AI systems can be recognized as inventors. However, in April 2022,
this decision was overturned. The main argument of the courts is that no AI
creates a new invention on its own, it is a tool that ultimately acts on human
instructions. However, Thaler and Abbott are not giving up and continue to
appeal the decisions. In connection with the latest refusal, Thaler asks for a re-
hearing at the Federal Circuit level. The current position on this issue is as
follows: at the current stage of legal development, the illegal behavior of
artificial intelligence must always result in human responsibility. Thus, the PACE
recommendations “Merging with Technology, Artificial Intelligence and Human
Rights” of April 28, 2017 No. 2102 also explicitly state that responsibility for the
actions of artificial intelligence lies with a person regardless of the
circumstances of the incident, and even references to the independence of
decisions made by artificial intelligence units cannot exempt their creators,
owners and operators from liability. Such decisions are a natural consequence of
the doctrinal approach to the mass adoption of artificial intelligence systems and
their assimilation into everyday life, which is being tested in modern conditions.
In this context, the role of international law may consist in coordinating
the development of legal regulation; possibly in developing internationally
agreed guidelines to ensure the integration of fundamental values in the
development of autonomous systems using artificial intelligence; in adapting
existing norms and concepts; in filling gaps in legal regulation; in developing and
adopting the concept of responsibility [7].
Without questioning the need for legal regulation of these technologies,
the scientific community is debating the scale of such regulation. In particular,
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
181
whether it is worth limiting ourselves to modernizing current legislative acts
(and perhaps their broad interpretation) or whether it is necessary to introduce
a new branch of legislation for this area. Representatives of the first approach
point to the need to avoid a situation of excessive regulation of technologies that
hinders their development. As an example, they refer to the history of the
formation of the Internet, the lack of regulation of which allowed it to fully
develop its potential (although it is worth noting that the history of the Internet
has apparently approached a point beyond which its development free from
regulation becomes impossible due to the clash of interests of transnational
technology corporations and states). Representatives of the opposite approach
justify their position by the specifics of artificial intelligence systems, which
require the adoption of new legislation. [3].
Thus, artificial intelligence-based technologies are already an integral part
of modern life, but the legal regulation of such developments still remains an
open question. It is no exaggeration to say that legal systems are currently
hopelessly behind the development of artificial intelligence technology. There is
no universal approach to the legal regulation of this problem, which is explained
by the conciliatory nature of international law, which dictates a slow and
painstaking process of decision-making and the creation of legal norms, which is
also reinforced by the divergence of interests of subjects of international law,
especially states. This is reflected at the national level of legal regulation. Despite
the relevance of existing and potential problems associated with activities in the
field of artificial intelligence, international norms, as well as the legislation of
individual states, currently do not have the necessary tools and mechanisms to
cope with the threats and risks posed.
References:
1.
Берггольц В. Правовой статус и разграничение ответственности при
разработке и использовании инструментов искусственного интеллекта //
International Journal of Humanities and Natural Sciences, vol. 6-3 (45), 2020. –
C.23-27.
2.
Васильев А. Искусственный интеллект: правовые основы. //Известия
АлтГУ. Юридические науки. 2018. №6 (104) – C.23-26.
3.
Казанцев
Д.
Анализ
тенденций
становления
правового
регулирования искусственного интеллекта и роботов // Научные записки
молодых исследователей № 1/2022. - С.66-76.
4.
Кашкин С. , Покровский А. Искусственный интеллект, робототехника
и защита прав человека в ЕС. // Вестник Университета им. Кутафина.
№4/2019. – C.64-90.
ACADEMIC RESEARCH IN MODERN SCIENCE
International scientific-online conference
182
5.
Латипова А. Искусственный интеллект и международное право.
https://inter-legal.ru/iskusstvennyj-intellekt-i-mezhdunarodnoe-pravo
6.
Понкин И.В., Редькина А.И. Искусственный интеллект с точки зрения
права. //Вестник РУДН. Серия: Юридические науки 2018 Т. 22. № 1. – C.91-
109.
7.
Шестак В.А., Волеводз А.Г. Современные потребности правового
обеспечения искусственного интеллекта: взгляд из России// Russian
Journal of Criminology, 2019, vol. 13, no. 2, pp. 197–206/ Всероссийский
криминологический журнал. 2019. Т. 13, № 2. C. 197–206.
8.
High-Level Expert Group on Artifcial Intelligence. (2019). A defnition of AI:
Main capabilities and scientifc disciplines. European Commission.
https://ec.europa.eu/newsroom/ dae/document.cfm?doc_id=56341
