Instituto de Estudios Políticos y Derecho Público "Dr. Humberto J. La Roche"

de la Facultad de Ciencias Jurídicas y Políticas de la Universidad del Zulia

Maracaibo, Venezuela

Esta publicación cientíca en formato digital es continuidad de la revista impresa

ISSN-Versión Impresa 0798-1406 / ISSN-Versión on line 2542-3185Depósito legal pp

197402ZU34

ppi 201502ZU4645

Vol.40 N° 72

Enero

Junio

2022

ISSN 0798- 1406 ~ De pó si to le gal pp 198502ZU132

Cues tio nes Po lí ti cas

La re vis ta Cues tio nes Po lí ti cas, es una pu bli ca ción aus pi cia da por el Ins ti tu to

de Es tu dios Po lí ti cos y De re cho Pú bli co “Dr. Hum ber to J. La Ro che” (IEPDP) de la Fa-

cul tad de Cien cias Ju rí di cas y Po lí ti cas de la Uni ver si dad del Zu lia.

En tre sus ob je ti vos fi gu ran: con tri buir con el pro gre so cien tí fi co de las Cien cias

Hu ma nas y So cia les, a tra vés de la di vul ga ción de los re sul ta dos lo gra dos por sus in ves-

ti ga do res; es ti mu lar la in ves ti ga ción en es tas áreas del sa ber; y pro pi ciar la pre sen ta-

ción, dis cu sión y con fron ta ción de las ideas y avan ces cien tí fi cos con com pro mi so so cial.

Cues tio nes Po lí ti cas apa re ce dos ve ces al año y pu bli ca tra ba jos ori gi na les con

avan ces o re sul ta dos de in ves ti ga ción en las áreas de Cien cia Po lí ti ca y De re cho Pú bli-

co, los cua les son so me ti dos a la con si de ra ción de ár bi tros ca li fi ca dos.

ESTA PU BLI CA CIÓN APA RE CE RE SE ÑA DA, EN TRE OTROS ÍN DI CES, EN

Re vicyhLUZ, In ter na tio nal Po li ti cal Scien ce Abs tracts, Re vis ta In ter ame ri ca na de

Bi blio gra fía, en el Cen tro La ti no ame ri ca no para el De sa rrol lo (CLAD), en Bi blio-

gra fía So cio Eco nó mi ca de Ve ne zue la de RE DIN SE, In ter na tio nal Bi blio graphy of

Po li ti cal Scien ce, Re vencyt, His pa nic Ame ri can Pe rio di cals In dex/HAPI), Ul ri ch’s

Pe rio di cals Di rec tory, EBS CO. Se en cuen tra acre di ta da al Re gis tro de Pu bli ca cio-

nes Cien tí fi cas y Tec no ló gi cas Ve ne zo la nas del FO NA CIT, La tin dex.

Di rec to ra

OIRALITH

M. C

HIRINOS

ORTILLO

Co mi té Edi tor

Eduviges Morales Villalobos

Fabiola Tavares Duarte

Ma ría Eu ge nia Soto Hernández

Nila Leal González

Carmen Pérez Baralt

Co mi té Ase sor

Pedro Bracho Grand

J. M. Del ga do Ocan do

José Ce rra da

Ri car do Com bel las

An gel Lom bar di

Die ter Nohlen

Al fre do Ra mos Ji mé nez

Go ran Ther born

Frie drich Welsch

Asis ten tes Ad mi nis tra ti vos

Joan López Urdaneta y Nil da Ma rín

Re vis ta Cues tio nes Po lí ti cas. Av. Gua ji ra. Uni ver si dad del Zu lia. Nú cleo Hu ma nís ti co. Fa-

cul tad de Cien cias Ju rí di cas y Po lí ti cas. Ins ti tu to de Es tu dios Po lí ti cos y De re cho Pú bli co

“Dr. Hum ber to J. La Ro che”. Ma ra cai bo, Ve ne zue la. E- mail: cues tio nes po li ti cas@gmail.

com ~ loi chi ri nos por til lo@gmail.com. Te le fax: 58- 0261- 4127018.

Vol. 40, Nº 72 (2022), 690-703

IEPDP-Facultad de Ciencias Jurídicas y Políticas - LUZ

Recibido el 15/09/2021 Aceptado el 21/12/2021

Legal regulation of articial

intelligence and robotic systems:

review of key approaches

DOI: https://doi.org/10.46398/cuestpol.4072.40

Dmitry Kuteynikov *

Osman Izhaev **

Valerian Lebedev ***

Sergey Zenin ****

Abstract

The aim of the article is to study various approaches to legal

regulation of AI articial intelligence and robotic systems in the

European Union, USA, and China. These regions are the world’s

largest centers of technological development and therefore each

of them has perfected a unique approach to legal regulation on

the limits, scopes, and proper uses of AI. His achievements are

widely used by other countries. The authors used the methods of

analysis of scientic documents, laws, and legal regulations. In addition,

this article reviews the basic conceptual approaches available in the world

for the formation of legal regulation in the eld of the use of AI and robotic

systems. It is concluded that policies regulating articial intelligence are

not limited to one area and, in general, are intended to protect the rights

and freedoms of citizens, regardless of the eld of application of AI in the

social order.

Keywords: articial intelligence; robotics; legal status; human rights;

technological innovations.

* Kutan Moscow State Law University, Moscow, Russia. ORCID ID: https://orcid.org/0000-0003-

1448-3085

** Kutan Moscow State Law University, Moscow, Russia. ORCID ID: https://orcid.org/0000-0003-

3777-8927

*** Kutan Moscow State Law University, Moscow, Russia. ORCID ID: https://orcid.org/0000-0002-

7642-1325

**** Tyumen State University, Tyumen, Russia; Kutan Moscow State Law University, Moscow, Russia.

ORCID ID: https://orcid.org/0000-0002-4520-757X

691

CUESTIONES POLÍTICAS

Vol. 40 Nº 72 (2022): 690-703

Regulación legal de la inteligencia articial y los

sistemas robóticos: revisión de enfoques clave

Resumen

El objetivo del artículo es estudiar varios enfoques de regulación legal

de la inteligencia articial IA y los sistemas robóticos en la Unión Europea,

EE. UU. y China. Estas regiones son los centros de desarrollo tecnológico

más grandes del mundo y, por lo tanto, cada una de ellas ha perfeccionado

un enfoque único para la regulación legal sobre los límites, alcances y usos

adecuados de la AI. Sus logros son ampliamente utilizados por otros países.

Los autores utilizaron los métodos de análisis de documentos cientícos,

leyes y reglamentos legales. Además, el artículo revisa los enfoques

conceptuales básicos disponibles en el mundo para la formación de la

regulación legal en el campo del uso de la IA y los sistemas robóticos. Se

concluye que las políticas que regulan a la inteligencia articial no se limitan

a un área y, en general, tienen como propósito proteger los derechos y

libertades de los ciudadanos, independientemente del campo de aplicación

de la IA en el orden social.

Palabras clave: inteligencia articial; robótica; estatus legal; derechos

humanos; innovaciones tecnológicas.

Introduction

The legal and technical regulation of autonomous technologies is being

discussed and developed. Many countries have adopted national strategies

for developing articial intelligence and robotics, which, among other

things, contain general approaches to regulating their use. Currently, the

legislators of many states who declare their intentions to become leaders in

this area face an important task of selecting the concept of legal regulation

to form their regulatory framework.

There are dierent approaches to regulation from the US, Europe, and

China. On the one hand, these regions are the world’s largest centers of

technological development. On the other hand, each of them has developed

a unique approach to the legal regulation of the above-mentioned relations.

1. Methods

Author study the AI legal regulation and robotic systems in certain

countries and regions (EU, USA, and China).

692

Dmitry Kuteynikov, Osman Izhaev, Valerian Lebedev y Sergey Zenin

Legal regulation of articial intelligence and robotic systems: review of key approaches

The rst paragraph considers the legal framework for regulating the use

of AI in EU countries. The main research object is the draft regulation “On

harmonized rules on AI (AI Act) and amending certain union legislative

acts”. The key idea of this document is to apply a risk-based approach in

regulation, whose essence is to classify AI-driven systems into dierent

categories depending on their potential threat to health, safety, and

fundamental human rights.

The second paragraph examines the specic regulation of public

relations and the use of AI systems in the US. Both at the federal and state

levels, there are no cross-cutting laws that form a unied legal approach to

regulating the area in question. This is a conscious legal policy based on the

principle of “ad hoc” regulation that presupposes the regulation of social

relations as they naturally develop.

The third paragraph dwells on the legal regulation of using AI in

China. The main framework of legal regulation in the country is the “New

Generation AI Development Plan” which denes the key goals and principles

for developing this sphere of public relations. Despite ambitious concepts

and the introduction of innovations, the development of legal regulation in

the eld of AI-driven and robotic systems is associated with political risks.

2. Results

EU countries are about to adopt a unied end-to-end regulatory act

that will aect dierent spheres of public relations. This regulation should

utilize a risk-based approach focused on a person. In this connection, the

main criterion is the potential threat posed by AI to individual rights. The

advantage of such a regulatory system is that it is not limited to one area and,

in general, aims at protecting the rights and freedoms of citizens regardless

of the eld of AI application. However, this also conditions a disadvantage

associated with the fact that an excessive number of regulatory requirements

for actors involved in the development, distribution, and use of AI-powered

systems can signicantly slow down the growth of the industry, which will

have a negative impact (at least in economic terms) on the quality of life.

In the US, there are no cross-cutting laws that form a unied legal

approach to regulating the area in question both at the federal and state

levels. This is a conscious legal policy of the state based on the principle

of “ad hoc” regulation that presupposes the regulation of social relations

as they naturally develop. Thus, the regulatory legal acts adopted in the

US are fragmented and not uniform from the semantic perspective. Legal

regulation is not complex and aects certain spheres (for example, transport

and medicine) or narrow issues (for instance, hiring discrimination) related

to the development and functioning of AI-powered systems. This approach

693

CUESTIONES POLÍTICAS

Vol. 40 Nº 72 (2022): 690-703

is benecial from an economic viewpoint since it does not imply the creation

of a massive mechanism of legal regulation. At the same time, there is a

risk that a signicant number of incidents in the eld of human interaction

with AI-driven systems will occur before a sucient regulatory framework

is formed in the US.

In China, legal regulation is based on the “New Generation AI

Development Plan” which denes the key goals and principles for

developing this sphere of public relations. This concept represents only a

general model and goals of future legal regulation. Accordingly, it should

be considered in connection with other regulatory legal acts. On the one

hand, China has adopted ambitious development concepts in the eld of

AI systems and robotics. This country introduces innovations into public

life much more actively than the EU and US due to centralized regulation.

On the other hand, their actual implementation is entrusted to private

companies and local authorities, and the state retains ample opportunities

for control in all spheres.

3. Discussion

3.1. Legal basis for regulating AI application in Europe: risk-

based approach

The EU is actively seeking its own way of developing AI (Straus, 2021).

The European Commission is committed to implementing a set of policies

aimed at stimulating the industry while respecting fundamental human

rights. This body put forward three proposals to transform Europe into a

space where innovations actively develop, ensure the safe use of technology,

and support a favorable business environment. These proposals include: a)

a legal framework for the regulation of AI; b) an approach to establishing

accountability for AI-related incidents (planned for release in the last

quarter of 2021 – rst quarter of 2022); c) updating industry-specic

legislation (e.g. safety regulations, the General Product Safety Directive –

Q2 2021).

In the course of the study, we paid special attention to the legal concept

embraced in a Regulation of the European Parliament and of the Council

laying down harmonized rules on AI (AI Act) and amending certain union

legislative acts issued in April 2021 (EUR-Lex, 2021). The key idea of

this document is to apply a risk-based approach in regulation. It aims at

classifying AI-driven systems into dierent categories, depending on their

potential threat to health, safety, and fundamental human rights. Thus, AI-

powered systems can be recognized as means that create: a) an unacceptable

risk; b) a high risk; c) a low risk.

694

Dmitry Kuteynikov, Osman Izhaev, Valerian Lebedev y Sergey Zenin

Legal regulation of articial intelligence and robotic systems: review of key approaches

AI-powered systems with an unacceptable risk are completely prohibited

because their use violates the universal values recognized by the EU. In

particular, the use of various systems that aect the person’s consciousness

against their will is not allowed, namely, various manipulative techniques

that address various groups of the population: children, seniors, persons

with mental disorders, etc.

If the previous category does not cause any controversy, the next one is

rather ambiguous. High-risk AI systems fall under a whole set of regulatory

requirements and are allowed on the European market if they fully comply

with them. The criteria for assigning a specic AI-driven system are their

functional characteristics and goals. Within this group, they are divided

into:

a. AI systems to be used as a safety component of products subject to

prior third-party conformity assessment.

b. AI systems, whose exploitation can aect the state of human rights

and whose list is indicated in a separate annex (for example, use

in law enforcement, the administration of justice, or the eld of

democracy).

Such requirements represent a system of continuous risk management:

monitoring, identifying, and assessing them with due regard to the

available technical capabilities; thorough testing of these systems during

development and before commission based on the purpose of a specic AI-

powered system. Particular attention should be paid to data processing, i.e.,

information should be up-to-date, representative, correct and complete.

Finally, the third category includes low-risk AI systems that do not need

any regulation. However, attention is drawn to the fact that responsible

actors might comply with codes of ethics when creating, developing, and

using such systems.

The use of AI systems raises legal issues at the level of national legislation

in European countries. These issues concern, inter alia, human rights,

condentiality, fairness, algorithmic transparency, and accountability

(Wachter et al., 2021). Many states emphasize the need to assess the

existing legal framework and enact new legislation to provide favorable

legal conditions for the successful implementation and operation of AI-

driven systems.

For example, Belgium adopted a Royal Decree on tests with automated

vehicles in March 2018 (Belgisch Staatsblad, 2018). In 2017, a similar act

was adopted by the Danish parliament that amended the road trac law to

allow tests of unmanned vehicles. In addition, Denmark has amended the

Danish Financial Statements Act which stipulates that the largest companies

adhering to data ethics policies must provide compliance information, while

695

CUESTIONES POLÍTICAS

Vol. 40 Nº 72 (2022): 690-703

companies that do not have a data ethics policy are required to explain why

they do not have such a policy.

In 2020, Finland adopted a new law aimed at developing the

smooth operation and safety of transport, creating the prerequisites for

digitalization and automation of road trac (Belgisch Staatsblad, 2020).

The Netherlands also adopted and implemented regulations on self-driving

vehicles, automated decision-making by law enforcement agencies, and

the prevention of discrimination in employment when using automated

systems (Government of the Netherlands, 2019). Finally, Lithuania passed

a law on autonomous driving which allows driving cars without humans on

board (Chancellery of the Seimas of the Republic of Lithuania, 2021).

Thus, countries start to develop industry-specic regulations for certain

areas of AI that are currently not covered or insuciently covered by the

existing EU legislation. In this context, it is worth mentioning unmanned

vehicle regulations. Many states have enacted regulations allowing the

testing of unmanned vehicles and related technologies on public roads.

Other signicant areas of regulation are data (namely, in health care) and

automated decision making. For example, Norway works on proposals to

amend its health register to distinguish between the use of data for patient

care and the rules for obtaining consent from individuals. Slovakia also

prepares a new Act on Data to better dene data protection rules, disclosure

principles, data access, and open data rules.

In addition, many European states consider creating special legal

regimes for experimentation with AI, for example, by developing regulatory

sandboxes. While several EU member states have announced it in their

national AI strategies, the overall development of regulatory sandboxes for

AI is still insucient.

Based on the foregoing, Europe is about to adopt a unied end-to-end

regulatory act that will aect dierent spheres of public relations. This

regulation should utilize a risk-based approach focused on a person. In

this connection, the main criterion is the potential threat posed by AI to

individual rights. The advantage of such a regulatory system is that it is

not limited to one area and, in general, aims at protecting the rights and

freedoms of citizens regardless of the eld of AI application. However, this

also conditions a disadvantage associated with the fact that an excessive

number of regulatory requirements for actors involved in the development,

distribution, and use of AI-powered systems can signicantly slow down

the growth of the industry, which will have a negative impact (at least in

economic terms) on the quality of life.

696

Dmitry Kuteynikov, Osman Izhaev, Valerian Lebedev y Sergey Zenin

Legal regulation of articial intelligence and robotic systems: review of key approaches

3.2. Use of AI systems in the US: specic regulation of public

relations

In the US of America, there has been an increase in the number of AI

regulations submitted to the federal legislature over the past few years.

Thus, the 115

Congress (2017-2018) received 50 bills mentioning AI,

and the 116th Congress (2019-2020) considered 175 (The United States

Congress, 2021). This indicates a signicant interest on the part of

legislators to regulate this area. However, only seven of such bills were

adopted and entered into force. In addition, these acts are not connected

and are concerned with such dierent issues as funding, the development

of scientic centers, defense, and international relations.

Currently, the sphere of AI is regulated by bylaws issued by executive

authorities. The rst document was the Executive Order of the President of

the US of February 11, 2019, No. 13859 “Maintaining American Leadership

in AI” (hereinafter referred to as the Order) (The White House, 2019). It

denes ve key principles to develop state policy in the eld of developing

AI. The Order formulates such principles of state policies as promoting the

introduction of technological breakthroughs; the development of technical

standards; the training of workers with skills in the development and

application of AI technologies; protecting the American values, including

civil liberties and privacy, increasing public condence in AI technologies;

developing an international environment to support US developments and

open up new markets with due regard to the need to conceal technological

advantages and critical AI-powered technologies from strategic competitors

and rival countries.

Based on this Order, the Administration of the President of the US

approved the Memorandum for the heads of executive departments and

agencies in November 2020 (The White House, 2020). This document sets

out the principles that should be followed by the executive authorities to

develop normative and non-normative approaches to the implementation

and operation of AI systems, both at general and sectoral levels. These

principles are laid down in the following manner:

• Creating public condence in AI systems.

• Public participation in decision-making.

• The use of the most objective and scientically grounded information

in the activities of executive authorities.

• The application of a risk-based approach in regulating the use of AI

systems.

• Considering the benets and costs of using AI systems.

697

CUESTIONES POLÍTICAS

Vol. 40 Nº 72 (2022): 690-703

• The use of result-oriented exible approaches to regulation.

• Analyzing possible discriminatory consequences from the use of AI

systems.

• Ensuring the transparency and accountability of decisions made by

AI systems.

• Securing the functioning of AI systems at all stages.

• Implementing interdepartmental coordination between various

government bodies.

Under this document, if the existing regulatory framework is sucient

for the use of a specic technological solution in the eld of AI, or the

development of new legislation is incommensurable with the predicted

economic benets of this solution, then its use should be stopped or

replaced with non-regulatory approaches, including guidelines or programs

for the implementation of public policies in certain sectors of the economy,

pilot programs and experiments, and voluntary consensus standards and

frameworks.

Following the entry into force of the President’s Order, the House of

Representatives adopted a Resolution “Supporting the development of

guidelines for ethical development of AI” (The United States Congress,

2019). It contains tasks correlating with the provisions of the Order,

whose implementation should be ensured: the interaction of industry,

government, academic community, and civil society; the transparency and

“expandability” of AI systems; empowering women and underrepresented

or marginalized groups of society; information condentiality and personal

data protection; career opportunities for dierent social classes; the

accountability and oversight of all automatic decision-making systems; life-

long education in engineering, social sciences, and humanities; equitable

access to technological services; the interdisciplinary research of safe and

useful AI; the security and control of AI systems.

At the state level, the most regulated aspect of AI is the use of unmanned

vehicles. More than half of the states have enacted legislation that, to one

degree or another, allows the use of such vehicles on public roads (National

Conference of State Legislatures, 2020).

Let us consider the existing laws in dierent states. Illinois passed

the AI Video Interview Act (Illinoys General Assembly, 2020), requiring

employers to notify interviewees that AI might be used to assess them.

Before the interview, they should get the consent of candidates, provide

information on how AI works and what criteria are used to assess their

professional suitability. Alabama has two laws that recognize the impact

of AI technologies on the growing number of jobs in the state (Waggoner,

2019a) and set up a state commission on AI to review and advise on all

698

Dmitry Kuteynikov, Osman Izhaev, Valerian Lebedev y Sergey Zenin

Legal regulation of articial intelligence and robotic systems: review of key approaches

aspects of developing and using AI in various spheres (Waggoner, 2019b).

California passed a law that requires each government agency to provide

information to the public prior to approving a subsidy for the development

of warehouse distribution centers, regularly report on the reduction or

replacement of jobs due to automation to the Governor’s Oce of Business

and Economic Development (Medina, 2019). By a 2019 law, the state of

New York established an interim state commission to study the regulation

of AI, robotics, and automation until December 2020 (Savino, 2019).

In 2021, more than 10 states introduced draft laws or parliamentary

resolutions related to the regulation of AI. All of them are under consideration

and are related to such issues as the development of AI (Alabama); the

use of methods that minimize the risk of adverse consequences caused

by the automated decision-making systems made by government bodies

(California); tax benets (Hawaii); establishing requirements for ensuring

the fairness and transparency of automated decision-making systems used

by government agencies, as well as the condentiality of consumer data

(Massachusetts); the verication of computer system algorithms and logical

formulas used by the unemployment agency (Michigan); the prohibition of

discrimination against certain categories of the population by automated

decision-making systems (New Jersey); the establishment of a commission

to oversee the impact of technology on the labor market and the state’s

economy as a whole (New York); the establishment of an advisory group

to eliminate bias in government software (Vermont); the development

of guidelines for public procurement and the use of automated decision-

making systems to protect consumers and increase market transparency

(Washington).