9.

The evolution of copyright in the age

of artificial intelligence: challenges
and perspectives
Ewa Milczarek

INTRODUCTION

The intensive development of artificial intelligence (AI) in recent years has

significantly affected various socioeconomic sectors and generated new ethical
and legal issues. This impact has not spared the creative sector. Some AI tools
assist or complement the human creative process, and can generate cultural
content without human input. This raises questions about the protection of
such works by copyright law.
In this context, the research conducted within the framework of the chapter
centres around exploratory questions:

1. Can AI creativity be treated as a ‘work’ within the meaning of copyright

law?
2. If the creativity generated by AI qualifies as a work of authorship, to
whom do the rights to this work belong?
3. Should the copyright protection system evolve to extend protection to AI
creations as well?

The chapter is divided into three main parts. The first part analyses the current
legal status in the context of the possibility of copyright protection of AI
works. The goal is to determine whether AI works meet the conditions for
being surrounded by copyright protection. Can AI be a legitimate subject?
In the second part, the prospect of the development of legislation related to
AI works is examined. An assessment is made as to whether and under what
conditions it is possible to protect it, and to whom the copyright in such crea-
tivity belongs. The last part presents the problem of copyright protection of the
source works on which artificial intelligence works.

158

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
 The evolution of copyright in the age of artificial intelligence 159

The research methods used in this chapter are the dogmatic-legal method,
the comparative-legal method, and case study analysis. The chapter is of an
overview nature.

COPYRIGHT OF ‘WORKS’ OF ARTIFICIAL

INTELLIGENCE – CURRENT STATUS

Generative AI refers to a type of artificial intelligence that can generate

human-like text and creative content (e.g. music and images), as well as
consolidate data from various sources for analysis1 (e.g., ChatGPT, Bard,
DALL-E, Midjourney, and DeepMind). These technologies build on the vast
number of pre-existing human-authored works and use inferences from these
works to generate new content. Some systems work in response to user textual
instructions, called prompts.2
Basically, we can deal with three situations related to the use of generative
AI in the context of the topic of the chapter:

1. the creation of a work by AI;

2. the creation of a work in which we can distinguish between the human-made
parts and those that have been created by AI;
3. creation of a work by a human with the help of AI.

The legal system is anthropocentric. The subjects of rights, including copy-

right, can only be human beings.3 The purpose of copyright law is to protect
creative, personal human activity. AI is not a subject of law, so no subjective
rights belong to it. This is in line with existing case law on other ‘non-human’
works (e.g., created by an animal).4
In relation to AI creativity, two points can be made in this regard. First, this
creativity is devoid of the ‘human factor’. An activity that merely imitates

1Marie T Dasborough, ‘Awe-inspiring advancements in AI: The impact of

ChatGPT on the field of organizational behavior’ (2023) 44 J. Organ. Behav 2,
177.
2 OECD, ‘Artificial Intelligence, Machine Learning and Big Data in Finance:
Opportunities, Challenges, and Implications for Policy Makers’ (2021) <https://​
www​.oecd​.org/​finance/​artificial​-intelligence​-machine​-learning​-big​-data​-in​
-finance​.htm> accessed 28 November 2023.
3 Bohdan Widła, ‘Prawa natury? Własność i podmiotowość prawna w antro-
pocenie’, in Katarzyna Jasikowska and Michał Pałasz (eds), Za pięć dwunasta
koniec świata. Kryzys klimatyczno-ekologiczny głosem wielu nauk (Uniwersytet
Jagielloński w Krakowie 2022) 519.
4 Naruto v Slater, 888 F.3d 418, 426 (9th Cir. 2018).

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
160 Artificial intelligence and international human rights law

someone else’s creativity, nature, or commonly used structures, and which

does not have the creative and personal character5 attributed to a human being,
is not a work. Not being created by a human being, it cannot be protected.
This position is consistent on both sides of the Atlantic. The EU legislature6
and jurisprudence7 directly indicate that ‘works’ generated by artificial
intelligence are not protected by copyright in the absence of human creative
choices. Similarly, courts in the United States8 and Australia9 confirm that
works generated entirely by artificial intelligence without human input are
not subject to copyright. The court’s position is in line with earlier guidance
from the US Copyright Office issued in March 2023,10 which affirmed that
copyright protects only materials that are the product of human creativity.
Second, AI’s creativity is based on the analysis of the input material and the
suggestions of the user. AI itself is therefore not ‘creative’ but ‘reproductive’.
At present, therefore, such ‘works’ are not covered by copyright protection in
favour of AI.
Referring to the second of the analysed cases, in which we are dealing with
a work that is created in part by a human and in part by AI, and these individ-
ual components of the work are separable (for example, an illustrated book in
which the text was created by a human and the illustrations by AI). In such
a case, such a work as a whole is a copyrightable work, but those component
parts that were created by AI are no longer (following the example given,
both the text and the selection and arrangement of text and images made by

5 Monika Brzozowska-Pasieka, Anna Kruszewska, Aleksandra Sewerynik,

Jakub Słupski, Karolina Maria Wojciechowska, Joanna Woźniak and Julia
Zasacka, ‘Prawo autorskie – podstawowe konstrukcje i pojęcia’ in Aleksandra
Sewerynik (ed), Prawo autorskie w instytucjach kultury (C.H. Beck 2019) 1;
Gerald Spindler, ‘Copyright Law and Artificial Intelligence’ (2019) 50 Copyright
Law and Artificial Intelligence 1049, doi:10.1007/s40319-019-00879-w.
6 European Commission, Directorate-General for Communications Networks,
Content and Technology, Study on copyright and new technologies – Copyright
data management and artificial intelligence, Publications Office of the European
Union, 2022, p. 21 <https://​op​.europa​.eu/​en/​publication​-detail/​-/​publication/​cc293085​
-a4da​-11ec​-83e1​-01aa75ed71a1/​language​-en> accessed 16 December 2023.
7 Case C-5/08 Infopaq International A/S v Danske Dagbaldes Forening [2009]
ECR 2009 I-06569.
8 Thaler v Perlmutter, 1:22-cv-01564-BAH (D.D.C. 2023).
9 Acohs Pty Ltd v Ucorp Pty Ltd [2010] FCA 577.
10 Federal Register, Vol 88, No 51, 16 March 2023, Rules and

Regulations.

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
 The evolution of copyright in the age of artificial intelligence 161

a human are protected, but not the images themselves generated by artificial
intelligence).11
In the third case, that is, works created by humans using AI, such a work
is, in principle, subject to legal protection. The participation of artificial
intelligence in the creation of a work varies, so it is important to indicate the
demarcation line between works created with the help of AI which are subject
to protection and those which are not.
A human being, if they do not have sufficient creative control over the ele-
ments of a work generated by artificial intelligence, is not the ‘author’ of those
elements within the meaning of copyright law.12 A protected work will not be
one that has been produced by machine or by a mere mechanical process that
proceeds randomly or automatically, without any creative input or intervention
by the human author.13 The traditional elements of authorship of a work must
be produced by a human,14 and the work must be a representation of its original
intellectual concepts.15
In conclusion, in the current state of the law, copyright protection is not
granted to AI creations. When AI is used only as a tool to assist the author in
the creation process, the norms that protect their copyright apply.

COPYRIGHT OF ARTIFICIAL INTELLIGENCE

‘WORKS’ – A PERSPECTIVE

In the preceding part of the chapter, arguments were given for unequivocally
not granting copyright protection to artificial intelligence creations. It is worth
noting, however, that the lack of such protection creates a number of possible
problems that could hinder the development of this field. It is clear that AI, as
it is not a subject of law, cannot have copyright assigned to it. However, in the
process of creating such a work, there are legal entities that will suffer from
the failure to provide such protection – from AI developers/programmers, to
entrepreneurs who use such software, to individual creators.

11 U.S. Copyright Office, Cancellation Decision re: Zarya of the Dawn

(Registration # VAu001480196), 21 February 2023 <https://​www​.copyright​.gov/​
docs/​zarya​-of​-the​-dawn​.pdf> accessed 12 December 2023.
12 ibid 9.
13 U.S. Copyright Office, Compendium of U.S. Copyright Office Practices sec.

313.2 (3d ed. 2021).

14 Copyright Registration Guidance: Works Containing Materials Generated

by Artificial Intelligence, 16 March 2023 at 16192 <https://​copyright​.gov/​ai/​ai​

_policy​_guidance​.pdf> accessed 18 December 2023.
15 111 U.S. 53, 56 (1884).

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
162 Artificial intelligence and international human rights law

Thus, we are faced with answering a fundamental question: whether the

copyright protection system should evolve to extend protection to AI creations,
or whether the status quo should be preserved. To answer this question, it is
necessary to go back to the goals and axiology of copyright protection and
verify which of the indicated solutions is in line with its spirit.
Copyright protects the rights of creators by encouraging them to create
and enrich the cultural heritage. Thus, copyright supports society by ensuring
cultural activities are viable, diversity is protected, and cultural heritage acces-
sible.16 Hence, a balance is struck between the rights of creators, who receive
benefits in the form of economic rewards, and society, which receives benefits
in the form of creative works that otherwise would not have been created or
distributed. The European Union establishes a high level of protection for
copyright holders, allowing them to receive appropriate remuneration for the
use of their works or other protected subject matter, particularly when they are
made public.17
AI’s activities undoubtedly have a positive impact on enriching cultural
heritage.18 Here an example is ‘The Next Rembrandt’, a painting created by
AI after analysing the work of this painter.19 Deprivation of any protection
is detrimental to the creators using this tool, and thus will reduce the social
benefits of access to culture. Depriving AI works of protection is therefore
contrary to the spirit of copyright protection. Combining cases and survey
reports shows that the current copyright system in protecting AI’s creative
innovations is inadequate.20 Social change can similarly break down existing
legal categories, and require theoretical reassessments.21 It is therefore neces-
sary to seek solutions to extend protection to this sector, which are at the same
time compliant with the norms of the legal system.

16 Simone Schroff, ‘The purpose of copyright—moving beyond the theory’

(2021) 16 JIPLP 11, 1262, doi:10.1093/jiplp/jpab130.
17 Pursuant to recitals 4, 9 and 10 of Directive 2001/29, Art. 17 Charter of

Fundamental Rights of the European Union (C 326).

18 Anjan Chatterjee, ‘Art in an age of artificial intelligence’ (2022) 13 Front

Psychol, doi:10.3389/fpsyg.2022.1024449.
19 See <https://​www​.vml​.com/​work/​next​-rembrandt> accessed 1 September

2023.
20 Jingyi Cui, ‘Copyright and AI: Are Extant Laws Adequate?’ in Bo Hu,

Yunni Xia, Yiwen Zhang, Liang-Jie Zhang (eds), Big Data (Springer 2022),
doi:10.1007/978-3-031-23501-6_9.
21 Margot E Kaminski, ‘Authorship, Disrupted: AI Authors in Copyright and

First Amendment Law’ (2017) 51 U.C. Davis L Rev 589.

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
 The evolution of copyright in the age of artificial intelligence 163

PROPOSALS FOR COPYRIGHT PROTECTION OF AI

WORKS

The perspective on the development of AI in the context of copyright leads

to the need to analyse a number of possible paths that a legislator can take,
as well as their implications. In this chapter, selected paths will be indicated.
The purpose of this analysis is to present possible legislative directions in the
field of AI copyright. The multitude of indicated development paths makes it
necessary to limit the consideration to selected ones. The analysis will allow
answering exploratory questions: is it possible to protect the copyright of AI
works and under what conditions? Who will be entitled to the rights to AI
works? The analysis will address the following topics:

1. transformative creativity;
2. copyright of AI programmers;
3. work made for hire model (WMFH) – a solution for legal entities;
4. copyright for AI users – solution for individuals.
Referring to the argument regarding the ‘reproducibility’ of artificial intelli-
gence works, it is worthwhile to lean into the issue of transformative creativity.
The concept of transformative works was born in the United States.22 These
works of art take previous works of art and transform them into something
new. The boundary between a transformative work and a derivative work can
be difficult to define. The sensitive issue is whether the material has been used
to create something new or merely copied verbatim into another work. Under
the US Copyright Act,23 four factors are considered in determining fair use:
(1) the purpose and nature of the use; (2) the nature of the copyrighted work;
(3) the amount and substantiality of the portion used; and (4) the effect of the
use on the potential market or value of the copyrighted work. In the Polish
doctrine, the main criterion distinguishing an inspired work from a derivative
work is such creative modification of the elements of the inspired work that the
character of the inspired work is determined by its own individual elements,
and not by the elements taken over.24 In the case of an inspired work (covered
by independent copyright), only the thread of someone else’s work is taken,
and a new work is created with the characteristics of independent creativity.25

22Campbell v Acuff-Rose Music, Inc., 510 U.S. 569 (1994).

23Copyright Law of the United States (Title 17) and Related Laws Contained
in Title 17 of the United States Code.
24 Polish Supreme Court, judgment of 10 July 2014 (I CSK 539/14).
25 Polish Supreme Court (Civil Chamber), judgment of 23 June 1972 (I CR

104/72).

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
164 Artificial intelligence and international human rights law

Artificial intelligence works in a similar way – it takes known works and

transforms them into a new creation. Thus, can the generation of content by AI
be considered a form of transformative creation, which is usually protected as
new creation based on existing works? Works created by AI can undoubtedly
constitute new and original works.
Creation by artificial intelligence and ‘traditional’ creation still share the
common goal of expanding cultural heritage, despite differences in the cre-
ative act. Also, since machines have become intelligent and are increasingly
capable of making creative, innovative choices based on opaque algorithms,
the ‘intellectual’ in ‘intellectual property’ turns out to be perplexing.26 In the
EU, there are proposals to consider granting copyright protection for the sake
of the creative result, rather than for the sake of the creative process.27
The question remains: to whom should the rights associated with this work
belong? It is necessary to assign their copyright to a human author. Here we
can point to three possible actors: end users, programmers, and technology
owners/licensees. As Kalin Hristov points out, it is necessary to consider the
overall social benefit of the copyright attribution process.28
Some sort of solution – giving authorship to the programmer – can be
found in British legislation29 (as well as in India, Ireland, and New Zealand).
This approach is best encapsulated in UK copyright law, section 9(3) of the
Copyright, Designs and Patents Act (CDPA),30 which states: ‘In the case of
a literary, dramatic, musical or artistic work which is computer-generated, the
author shall be taken to be the person by whom the arrangements necessary for
the creation of the work are undertaken.’ There, the legal system establishes an
exception to any requirement for human authorship by recognizing the work
put into creating a program capable of generating works, even if the creative
spark is taken by a machine.31 The World Intellectual Property Organization

26 Jyh-An Lee, Reto M Hilty, and Kung-Chung Liu ‘Roadmap to Artificial

Intelligence and Intellectual Property’ in Jyh-An Lee, Reto M. Hilty, and
Kung-Chung Liu (eds), Artificial Intelligence and Intellectual Property (OUP
2021).
27 European Parliament Resolution on intellectual property rights for the devel-

opment of artificial intelligence technologies, 20 October 2020, 2020/2015(INI).

28 Kalin Hristov, ‘Artificial Intelligence and the Copyright Dilemma’ (2017)

57 IDEA: The IP Law Review 3, 443.

29 Kanchana Kariyawasam, ‘Artificial intelligence and challenges for copy-

right law’ (2020) 28 Int J Law Inf Technol 4, 279, doi:10.1093/ijlit/eaaa023.

30 Copyright, Designs and Patents Act 1988 c. 48.
31 Andres Guadamuz, ‘Artificial intelligence and copyright’, (2017) 5 WIPO

Magazine 14, 1 <https://​www​.wipo​.int/​export/​sites/​www/​wipo​_magazine/​en/​pdf/​

2017/​wipo​_pub​_121​_2017​_05​.pdf> accessed 28 November 2023.

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
 The evolution of copyright in the age of artificial intelligence 165

proposes to use this solution in other jurisdictions as a response to the AI

copyright challenge.32 However, this is not an adequate solution in every case.
It will limit the use and protection of AI creations by those using the software.
This solution seems appropriate when programmers are also users of AI.
Shlomit Yanisky-Ravid’s33 proposal, which creates the possibility of cop-
yright protection for businesses using AI, is to apply and adapt the WMFH
model. This includes the assumption that the autonomous AI system (like the
worker in the WMFH model),is the creator of the work. On the grounds that
the AI system is seen as a creative employee or autonomous user contractor,
the WMFH doctrine grants employers or individual principals copyright over
works of authorship created by employees or contractors. Under this model,
ownership, but also the duty to control and the question of liability, would
be assigned to those legal entities (individuals or organizations) that use and
benefit from AI systems. The policy rationale behind WMFH is to incentivize
employers and grant them control over works made on their behalf. The work
made for hire doctrine is a more fitting framework within which to situate
the problem of AI authorship because it represents an existing mechanism
for directly vesting ownership of a copyright in a legal person who is not the
author-in-fact of the work in question.34 It is no less necessary to adapt this
model to AI terms, including a reinterpretation of the terms ‘employee’ and
‘employer’ or ‘work for hire’. According to the new interpretation of the terms
in the WMFH doctrine, the authorship of works generated by artificial intelli-
gence will be granted to programmers and owners of AI devices.35
In the case of individuals using AI software, the European Union proposes
to assess the desirability of granting the copyright to such a ‘work’ to an indi-
vidual (including for commercial purposes) who edits and publishes it legally,
provided that the creators of the underlying technology have not objected to
such use.36 Also, in the case of AI consumer products, the copyright should
belong to the consumer rather than the AI producer because the end user has
final control over what the AI creates.37

32 ibid.
33 Shlomit Yanisky-Ravid, ‘Generating Rembrandt: Artificial Intelligence,
Copyright, and Accountability in the 3a Era—The Human-Like Authors are
Already Here—A New Model’ (2017) Mich State Int Law Rev 659.
34 Annemarie Bridy, ‘Coding Creativity: Copyright and the Artificially

Intelligent Author’ (2012) STLR 5, 12.

35 Hristov (n 28) 453.
36 European Parliament Resolution (n 27).
37 Zack Naqvi, ‘Artificial Intelligence, Copyright, and Copyright Infringement’

(2020) 24 Marq Intellectual Property L Rev 1, 50.

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
166 Artificial intelligence and international human rights law

As demonstrated in this chapter, solutions for providing protection to AI

works are feasible. As shown, determining to whom the rights to these works
would accrue is a complex task. A starting proposition should be to correlate
the rights to ‘works’ to technology developers. However, they should be able
to transfer rights in licensing agreements to users (private and corporate).
This is a solution that takes into account the interests of all stakeholders
while remaining in the spirit of protecting the rights of creators and the right
to dispose of the fruits of their labour. All of this is aimed at creating a legal
framework that will promote the further development of generative artificial
intelligence.

PROTECTING THE RIGHTS OF CREATORS OF

SOURCE WORKS

Consideration of the relationship between copyright and artificial intelligence

cannot take place without touching on the subject of copyright protection for
creators of source works. This problem is central to the development of arti-
ficial intelligence. Generative artificial intelligence operates on huge datasets
given to it as input. Copyright issues for these datasets can be complicated,
especially in situations where the data contains content created by many dif-
ferent creators.
Among the datasets on which AI works are public domain data and original
works by others, protected by copyright laws and regulations. Using the latter
to teach AI in terms of legal aspects may require obtaining appropriate approv-
als or entering into licensing agreements. Restricting access to this data would
have a significant impact on the efficiency of the systems. Also, access to large
datasets has become a competitive advantage for incumbents that can hinder
innovation and competition. Ensuring fair and open access to training data is
critical to the development and deployment of AI technology.38
In the European Union, copyright protects intellectual property until 70
years after the author’s death, or 70 years after the death of the last surviving
author in the case of a work of joint authorship39 (under the Berne Convention,40
the protected period is 50 years). During this time, the author has both moral
and economic rights (droit d’auteur). Economic rights guarantee, e.g., an

38 Nicola Lucchi, ‘ChatGPT: A Case Study on Copyright Challenges for

Generative Artificial Intelligence Systems’ (2023) Eur J Risk Regul 1, doi:10.1017/
err.2023.59.
39 Päivi Hutukka, ‘Copyright Law in the European Union, the United States

and China’ (2023) 54 IIC 1044, doi:10.1007/s40319-023-01357-0.

40 Berne Convention for the Protection of Literary and Artistic Works (1886).

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
 The evolution of copyright in the age of artificial intelligence 167

author control over their work and remuneration for its use through selling or
licensing. Moral rights give the right to refuse a modification of an author’s
work (right of integrity). Thus, the use of copyrighted works should be with
the consent of the authorized entity. It is therefore necessary to conclude an
agreement with the author, delineating the scope of data usage, establishing
limitations, specifying required permissions, and arranging the necessary
licences for using copyrighted material in AI training processes.41
Also, it is worth pointing out that most machine learning systems copy works
not to consume the expression copyright law protects, but to get access to the
facts or structures copyright law dedicates to the public.42 The works created
as a result of the process are based on a huge amount of data, in effect creating
a work in its own right. This process is closer to inspiration than to dependent
works. The introduction of a technologically neutral, open-ended exception
(akin to the fair use doctrine in the USA) could be explored.43 Permitted use is
a peculiar case of limiting the content of an author’s copyright.
Currently, there is no information on where generative AI gets its output
from. One can assume with some certainty that the data it uses includes copy-
righted data. Some artists, writers and programmers strongly object to the use
of their works as training data44 for generative AI systems and outputs that may
compete with or supplant their works. Lack of recognition and compensation
for the use of their original works are other sources of irritation for critics of
generative artificial intelligence.45 Therefore, users must be cautious and make
sure that generative artificial intelligence has been used in a lawful manner,
so that the content it generates does not infringe copyright.46 It is necessary to

41 Mauritz Kop, ‘The right to process data for machine learning purposes in the
EU’ (2021) 34 Harv J L& Tech 1.
42 Mark A Lemley and Bryan Casey, ‘Fair Learning’ (2021) 99(4) Tex Law

Rev 786.
43 Martin Kretschmer, Bartolomeo Meletti and Luis H Porangaba, ‘Artificial

intelligence and intellectual property: copyright and patents—a response by the

CREATe Centre to the UK Intellectual Property Office’s open consultation’
(2022) 17(3) JIPLP 321, doi:10.1093/jiplp/jpac013.
44 Edwards Benj, ‘Artists file class-action lawsuit against AI image gen-

erator companies’ (2023) Ars Technica <https://​arstechnica​.com/​information​

-technology/​2023/​01/​artists​-file​-class​-action​-lawsuit​-against​-ai​-image​-generator​
-companies/​> accessed 9 January 2024.
45 Pamela Samuelson, ‘Generative AI meets copyright’ (2023) 381(6654)

Science 158, doi:10.1126/science.adi0656.

46 John V Pavlik, ‘Collaborating with ChatGPT: Considering the implications

of generative artificial intelligence for journalism and media education’ (2023)

78(1) JMCE 84, doi:10.1177/10776958221149577.

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
168 Artificial intelligence and international human rights law

consider the design and implementation of guidelines and regulations for the
proper use of generative artificial intelligence47 aimed at preserving a sustain-
able creative ecosystem.
The European Union has begun work on regulating the operation of artificial
intelligence. It plans to impose transparency obligations on providers of such
services, including an obligation to disclose that content has been generated
by artificial intelligence; to prevent the model from generating illegal content;
and to publish the database used by the model. The European Commission is
working to create a general legal framework for the management of common
European data areas, particularly with regard to access to sensitive databases,
such as sensitive health data.48
In this regard, copyright protection is important to encourage source devel-
opers to make their content available, potentially accelerating the development
and improvement of generative AI models and promoting innovation in the
field. But at the same time, care must be taken to balance copyright protection
with access to training data to ensure that AI is developed responsibly and
in compliance with copyright laws. A solution may be to use the US fair use
doctrine while paying compensation to developers and users of AI technology
for traditional creators along the lines of copyright levies.

CONCLUSIONS

The development of generative artificial intelligence creates a number of chal-

lenges for copyright protection that require the intervention of the legislator.
As AI becomes an increasingly versatile and influential technology, new and
complex challenges arise in protecting creators’ rights and defining intellectual
property.
The results of the analysis allow us to state clearly that AI’s work is currently
not subject to copyright protection. Due to the negative impact of the lack of
such protection on the creative market, it was indicated that it is necessary to
take action to regulate the issue of AI works. It is crucial to adapt the current
copyright regulations to the new reality, taking into account the creative result
of AI, and not only the creation process. This work proposes treating AI works
as transformative creativity.

47 Fiona Fui-Hoon Nah, Ruilin Zheng, Jingyuan Cai, Keng Siau and Langtao
Chen, ‘Generative AI and ChatGPT: Applications, challenges, and AI-human col-
laboration’ (2023) 25(3) J Inf Technol Case Appl Res 277, doi:10.1080/15228053
.2023.2233814.
48 European Parliament Resolution (n 27).

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
 The evolution of copyright in the age of artificial intelligence 169

Additionally, consideration is given to copyright issues for works created

by AI which may be difficult to classify in a traditional copyright context. It
has been clearly demonstrated that AI works cannot be an authorized entity
and it is needed to assign their copyright to a human author. It was proposed
that the basic authorized entity would be AI programmers, who could transfer
rights to users – both individual and corporate – in software licence agree-
ments. Enterprises using generative AI in their operations could benefit from
a transformed and AI-adapted WMFH model, so that copyright remains within
the enterprise.
A key point discussed in the chapter is the issue of copyright for AI ‘train-
ing’ works. We discussed problematic issues related to identifying the creators
of source works in the case of AI, especially in situations where the data comes
from many different sources or is publicly available. The importance of licence
agreements and legal regulations in regulating the relations between the crea-
tors of source data, AI operators, and the creators of new works generated by
AI was also mentioned. A solution was also discussed in the form of applying
the American doctrine of fair use and correlating it with copyright levies for
the creators of source works.
The development of AI brings opportunities, but also threats. Too drastic
regulation may limit its development and thus negatively affect the compet-
itiveness of the economy. On the other hand, the lack of regulations may
generate problems related to lack of respect for fundamental rights. Therefore,
the issues discussed in the chapter emphasize the need to take into account new
challenges and the need to develop a clear, flexible, and fair legal framework
that will serve both creators and the development of artificial intelligence, and
so will contribute to further progress and the harmonious coexistence of these
two areas in an era of digital development. Copyright law in the context of AI
is an area that is still evolving, and legislative decisions and regulations must
keep track of the developing technology.
The purpose of copyright regulations is to protect and promote art and crea-
tivity. Regulation regarding AI in this sector must also provide for the impact
on ‘human’ creators. There is no doubt that the widespread use of AI will
have a revolutionary impact on the creative market.49 These transformations
might have negative impacts on the traditional creative sectors. Therefore, it is
important to conduct observations and, if necessary, introduce protections for

49 European Commission, Directorate-General for Communications Networks,

Content and Technology, Study on copyright and new technologies – Copyright data
management and artificial intelligence (Publications Office of the European Union
2022) 21 <https://​op​.europa​.eu/​en/​publication​-detail/​-/​publication/​cc293085​-a4da​
-11ec​-83e1​-01aa75ed71a1/​language​-ens> accessed 22 January 2024.

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/
170 Artificial intelligence and international human rights law

individual creators. In the long run, it should be remembered that AI technol-

ogy should be developed in a way that puts humans at the centre, and its appli-
cations should comply not only with the law, but also with ethical principles
and ensure that their applications do not cause unintended harm.50

50 Communication from the Commission to the European Parliament, the

Council, the European Economic and Social Committee and the Committee of
the Regions, ‘Building Trust in Human-Centric Artificial Intelligence’ Brussels 8
April 2019, COM(2019) 168 final.

Ewa Milczarek - 9781035337934

Downloaded from https://www.elgaronline.com/ at 11/08/2024 10:02:35AM
via Open Access. This work is licensed under the Creative Commons
Attribution-NonCommercial-No Derivatives 4.0 License
https://creativecommons.org/licenses/by-nc-nd/4.0/