Elaborating a Human Rights friendly Copyright Framework for Generative AI

Christophe Geiger

Abstract:

As works are increasingly produced by machines using artificial intelligence (AI) systems, with a
result often difficult to distinguish from that of a human creator, the question of what should be the
appropriate response of the legal system and, in particular, of the copyright system has become
central. If the creative input of the author has traditionally been the generator of copyright
protection, AI forces to reassess what in the creative process is special in human creativity and
where the creative input lies in AI-generated works. But it also poses more fundamental questions
on what the copyright system should achieve and who/what it should protect. In particular, as many
human authors will potentially face the competition of these AI machines on the market, new ways
of remunerating creators have to be imagined while making sure that the copyright system does
not stand in the way of these important technological developments.

This contribution analyses the copyright issues related to so-called “generative AI” systems and
reviews the arguments currently advanced to change the copyright regime for AI-generated works.
To do so, the underlying human rights framing intellectual property laws are used as the starting
point from which a balanced copyright framework for generative AI could (and even should) be
derived. It follows from the applicable human rights framework for copyright but also from the
anthropocentric approach of human rights that the protection of creators and human creativity
must be considered the point of reference when assessing future reforms with regard to copyright
and generative AI systems. This approach establishes generative AI systems as an instrument of


Christophe Geiger is a Professor of Law at Luiss Guido Carli University, Rome (Italy) and President of the
International Association for the Advancement of Teaching and Research in Intellectual Property (ATRIP); Global
Hauser Visiting Faculty, New York University School of Law, US (Fall 2023); Spangenberg Fellow in Law &
Technology at the Spangenberg Center for Law, Technology & the Arts, Case Western Reserve University School of
Law (Cleveland, US). Earlier versions of this article were presented in September and October 2023 in the context of
work-in-progress workshops at the Engelberg Center on Innovation law and Policy NYU School of Law; at the
Spangenberg Center for Law, Technology & the Arts at Case Western Reserve University School of Law (Cleveland),
and at American University’s Program on Information Justice and IP, American University Washington College of
Law (Washington D.C.). The author would like to thank the organizers and participants of these events, for their
invitations and helpful comments. The author is also very thankful to his research assistants at the NYU School of
Law, Nyusha Shafie and Raphael Weiss, for their great editorial assistance and their enthusiasm during the elaboration
process of the article.

1
the human creator – and not as a substitute. It also reinforces that copyright should be a tool to
protect creativity and creators, not a legal mechanism to secure the amortization of economic
investments in AI technology.

As a consequence, it is argued that the copyrightability of AI-generated outputs should be

considered with utmost care and only when AI is used as a technical tool for creators in their
creation process- meaning when they can serve a human author. At the same time, AI systems are
here to stay, and their development should not be inhibited as they can have many beneficial
aspects (including for creators) if appropriately regulated. The right to train generative AI systems
via machine learning technology can be derived from the right to science and culture and freedom
of (artistic) expression (Arts. 19 UDHR, 27.1 UDHR, 15.1 a and b ICESCR; 11 and 13 EUCF) as
AI can lead to useful advancements in science and arts; moreover, it is important for human
creators to be able to use outputs produced by generative AI in their creative process. This
grounding is even stronger when the training is conducted for research purposes as the training
process can then benefit also from the fundamental right to research-justification.

However, since a large amount of copyrighted works is required for the training of generative AI
systems, a remuneration obligation for these uses arises from a human rights perspective, in
particular when AI systems have a commercial purpose. It follows from the right to the protection
of moral and material interests of the creator (Arts. 27.2 UDHR, 15.1c) ICESCR; 17.2 EUCF, 1
Protocol No. 1, 8 ECHR) that authors must be adequately remunerated for the commercial use of
their works unless there is a strong justification legitimizing the use. For this reason, it is proposed
that the machine learning process using copyright-protected works to train the AI gives rise to a
limitation-based remuneration right to the benefit of human creators.

The article also briefly explores if and when the moral interest of creators deriving from human
rights protection could justify that they oppose the use of their work for training purposes of AI
systems. It is argued that the weaker the fundamental rights claim to train the AI is, the stronger
the moral rights claim could be. For example, training an AI to produce works for discriminatory
or racist purposes will benefit from a weaker (if any) fundamental rights protection, but will
potentially raise important moral concerns of the author of the work used for training purposes.

More generally, the article concludes that in order to secure a vibrant space for culture and
creativity, (finally) cherishing and putting the Human Author at the center of the copyright system

2
is necessary (and not only to built-up fences to the benefit of copyright industries). In doing so, it
might be able to have in the future AI-systems that serve creators and creativity, and not the other
way around.

I. Introduction

“Do you believe in the human heart? I don’t mean simply the organ, obviously. I’m speaking in
the poetic sense. The human heart. Do you think there is such a thing? Something that makes each
of us special and individual?” This is a question put to Klara, the narrator of Kazuo Ishiguro’s novel
Klara and the Sun, who is an “artificial friend”—an artificial intelligence (AI)-operated android-
that, in a not-too-distant future imagined by the Nobel Prize winner for literature, is meant to
replace companions for children1. This philosophical question also lies at the heart of the question
of the protectability by copyright of AI-generated outputs by generative AI systems: is there
something in the human creative process that makes it unique and different from any output
generated by a machine? As more and more works are produced by machines using AI, with a
result often difficult to distinguish from that of a human creator, the question whether the creative
input of the author- this moment of creative genius generated by the human mind, guided by
intuition and inspiration- should be the generator of copyright protection has become central. It
requires the identification and definition of what is an act of creativity2, but also poses more
fundamental questions on what the copyright system should achieve. Does the copyright system
protect authors -human authors- and remunerates them as a counterpart for their contribution to
collective cultural enrichment generated by their creations, or is copyright intended as incentive to
invest in the process of cultural production, no matter how the work has been generated?3

1
Kazuo Ishiguro, “Klara and the Sun”, Faber and Faber, 2021.
2
For a reflection on this issue, see e.g. C. Craig and I. Kerr, “The Death of the AI Author”, Ottawa L. Rev. 2020,
Vol. 52, Issue 1, p. 31. However, some advocates of protecting AI-generated creations through copyright law have
pointed out that the notion of creativity has not yet been clearly identified; for example, according to R. Abbott and E.
Rothman, it is problematic to consider that only human beings can “create” and thus draw legal consequences from it
(“Disrupting Creativity: Copyright Law in the Age of Generative Artificial Intelligence”, Florida Law Review
(forthcoming), p. 30, available on SSRN: https://ssrn.com/abstract=4185327: “There is no scientific, or even
philosophical, consensus on the nature of creativity. Without a clear understanding of creativity and thus what the
difference is between what an AI and a human being are doing, it seems problematic to argue that only what people
are doing counts as creative - and even more problematic to base laws on that assumption”).
3
We have repeatedly (critically) discussed these rationales and the deplorable trend of IP rights to become investment-
protection mechanisms, see e.g. C. Geiger, “Copyright as an Access Right, Securing Cultural Participation through the
Protection of Creators’ Interests”, in: R. Giblin and K. G. Weatherall (eds.), “What if we could reimagine copyright?”,
Australian National University (ANU) Press, 2016, p. 73 (at 74); C. Geiger, “Intellectual Property and Investment
Protection: A Misleading Equation”, in: V. Fischer, G. Nolte, M. Senftleben and L. Specht-Riemenschneider (eds.),
“Gestaltung der Informationsrechtsordnung - Festschrift für Thomas Dreier zum 65. Geburtstag”, C.H. Beck, 2022,

3
The issue is far from purely theoretical. It was at the root of a now famous dispute in the United
States between Dr. Stephen Thaler4 and the U.S. Copyright Office, with which he wanted to register
a work entitled “A Recent Entrance to Paradise” produced independently by an AI called
“DABUS” of which he is the owner, user and designer. Having had his application rejected by the
Copyright Office for the first time in 2019, he filed a petition for reconsideration of his application
with the same Office5. In its decision of February 14, 2022, the Office upheld the refusal to register
the work produced by the AI, pointing out that “only a human being can be considered an “author”
under U.S. copyright law, this quality being a prerequisite for the protection and registration of the
work”6. Dr. Thaler challenged that denial and filed a complaint to the Federal District Court7. His
motion for summary judgment seeking to register a copyright for the AI-generated work was denied
on 18 August 2023, the Court confirming “that the Copyright Office acted properly in denying
copyright registration for a work created absent any human involvement”8. This decision could be
the first episode in a long judicial journey, as it is likely that other courts will be seized in several

p. 7. And in the context of investment law, see e.g. C. Geiger “Regulatory and policy issues arising from intellectual
property and investor-state dispute settlement in the EU: A closer look at the TTIP and CETA”, in: C. Geiger (ed.),
“Research Handbook on Intellectual Property and Investment Law”, Edward Elgar, 2020, p. 505; “Excluding
Intellectual Property from Bilateral Trade and Investment Agreements: A Lesson from the Global Health Crisis”, in:
S. Frankel, M. Chon, G. Dinwoodie, J. Schovsbo and B. Lauriat (eds.), “Improving Intellectual Property: A Global
Project”, Edward Elgar, 2023, p. 426.
4
Dr. Thaler has initiated numerous lawsuits also seeking to have an AI recognized as an inventor under patent law.
For further references, see R. Abbott, “Intellectual Property and artificial intelligence, an introduction”, in R. Abbott
(ed.), “Research Handbook of Artificial Intelligence and Intellectual Property”, Edward Elgar, 2022, p. 18.
5
On the details of the case, see R. Abbott and E. Rothman, “Disrupting Creativity: Copyright Law in the Age of
Generative Artificial Intelligence”, supra note 2 at p. 15.
6
“Human authorship is a prerequisite to copyright protection in the United States and that the Work therefore cannot
be registered”, https://www.copyright.gov/rulings-filings/review-board/docs/a-recent-entrance-to-paradise.pdf;
similar refusals of registrations for AI generated works have since been issued on the same ground by the US Copyright
Office. See for example the decision of 21 February 2023, “Zarya of the Dawn”, deciding that the text of a graphic
novel (as well as the selection, coordination, and arrangement of the novel written and visual elements) could be
registered as protected by copyright law, but not the illustrations that were generated by an AI (Midjourney), as “the
images are not the product of human authorship” (https://www.copyright.gov/docs/zarya-of-the-dawn.pdf). See also
the decision by the Review Board of the USCO on 5 September 2023, which confirmed the refusal to register the work
generated by Midjourney “Théâtre D’opéra Spatial” (SR No. 1-11743923581) as the result was too much “machine”
and not enough “human”, despite the applicant having argued to have input at least 624 prompts and reworked the
output of Midjourney to reach the final result. For a critical comment, see E. Lee, “Prompting Progress: Authorship in
the Age of AI”, Florida Law Review 2024, Vol. 76 (forthcoming), available on SSRN:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4609687, arguing that the US Copyright Office practice
excluding AI-generated works from copyright registration is wrong.
7
DDC 2 June 2022, 1:2022cv01564 – “Thaler v. Perlmutter et al.”
8
United States District Court for the District of Columbia, 18 August 2023, Civil Action No. 22-1564 (BAH).
According to District Court Judge B. A. Howell, “Copyright has never stretched so far, however, as to protect works
generated by new forms of technology operating absent any guiding human hand, as plaintiff urges here. Human
authorship is a bedrock requirement of copyright”.

4
jurisdictions with similar requests aimed at recognizing the authorship of an AI, obviously with the
goal of vesting its owner (or the economic actors who designed, developed and/or financed it) with
ownership of the said right9. Of course, copyright regimes differ on both sides of the Atlantic, but
it is worth asking how such a dispute would be resolved in Europe as it is likely that similar
questions will soon be taken to court in the EU as well. Moreover, legislators on both sides of the
Atlantic are already pressed to extend copyright protection or other legal tools based on exclusivity
to AI generated output and it is likely that reform proposals will be put forward in the near future10.
When designing this new copyright framework for generative AI, legislators can find guidance in
the underlying human rights norms and their arising principles should lead the way11.

II. The Human Rights Framework for Copyright and Generative AI

Human rights are included in international and regional agreements as well as in national
constitutions. They bind the legislature as they rank very high in the hierarchy of norms, forming
the roots of positive law12; thus, human rights have to be considered by lawmakers and judges when
conceiving any legal framework applicable to a new technological, social or economic situation.
Rooted in ethical principles, fundamental rights offer possibilities for a balanced development of
intellectual property law. In fact, the rise of the use of human rights and fundamental rights in
shaping and using intellectual property norms has led in the last 20 years or so to a
“constitutionalization” of intellectual property law13, helping to design most recent evolutions in

9
The question of ownership of the right is fascinating but distinct from the question of potential protection. Indeed,
before defining who can be the owner of the copyright on this production, it is necessary to establish whether or not a
work generated by an AI can in itself benefit from copyright protection. This is the question that was repeatedly put to
the U.S. Copyright Office and on which we will focus here.
10
For OECD´s policy consideration on generative AI, see P. Lorenz, K. Perset and J. Berryhill, “Initial policy
considerations for generative artificial intelligence”, OECD Artificial Intelligence Papers, 2023, Issue 1, available at
https://doi.org/10.1787/fae2d1e6-en. See also the ongoing work of the Commissioner for Human Rights of the Council
of Europe in ensuring that human rights are strengthened and not undermined by artificial intelligence in her dialogue
with national authorities, national human rights structures and AI actors in general, available at
https://www.coe.int/en/web/commissioner/thematic-work/artificial-intelligence.
11
On the human rights implications of AI more generally, see A. Quintavalla and J. Temperman (eds.), “Artificial
Intelligence and Human Rights”, OUP (forthcoming).
12
See T. Mylly, “Intellectual Property and Fundamental Rights”, in: N. Bruun (ed.), “Intellectual Property beyond
Rights”, Helsinki, WSOY, 2005, at p. 187 sq., underlining that fundamental rights “provide the basic set of the most
fundamental norms and principles to which all areas of law are connected. They thus play a particular role in the pursuit
of coherence. Accordingly, private law and fundamental rights should be seen in a dialogical relationship: rather than
eliminating choice, autonomy and experimentalism, such a dialogue enables the realisation of certain basic values”.
13
C. Geiger, “Constitutionalizing’ Intellectual Property Law? The Influence of Fundamental Rights on Intellectual
Property in Europe”, IIC 2006, Vol. 37, Issue 4, p. 371; “Fundamental Rights as Common Principles of European
(and International) Intellectual Property Law”, in: A. Ohly (ed.), “Common Principles of European Intellectual
Property Law”, Mohr Siebeck, 2012, p. 223; “Reconceptualizing the Constitutional Dimension of Intellectual

5
the digital environment under the heading of what is increasingly understood as “digital
constitutionalism”14. Despite perfectible15, the constitutionalization of IP law has brough major
advancements for a balanced conceptualization and application of IP law and still can offer a useful
framework for the major future developments in the field, such as for example platform regulation,
access to information and research online or artificial intelligence16. Moreover, it offers a
transparent and workable methodology to mitigate conflicts of rights and secures a perspective on
innovation law rooted in the core values of the legal systems17, a field too often subject of influence

Property- An Update”, in: P. Torremans (ed.), “Intellectual Property and Human Rights”, 4 th ed., Kluwer Law
International, 2020, p. 117; J. Griffiths, “Constitutionalising or Harmonising? The Court of Justice, the Right to
Property and European Copyright Law”, European Law Review, 2013, p. 65; J. Griffiths, “Taking Power Tools to the
Acquis – The Court of Justice, the Charter of Fundamental Rights and European Union Copyright Law”, in: C. Geiger,
C.A. Nard and X. Seuba (eds.), “Intellectual Property and the Judiciary”, Edward Elgar Publishing, 2018, p. 144; T.
Mylly, “The Constitutionalization of the European Legal Order: Impact of Human Rights on Intellectual Property in
the EU”, in: C. Geiger (ed.), “Research Handbook on Human Rights and Intellectual Property”, Edward Elgar
Publishing, 2015, p. 103; J. Schovsbo, “Constitutional Foundations and Constitutionalization of IP Law – A Tale of
Different Stories?”, Zeitschrift für Geistiges Eigentum/Intellectual Property Journal, 2015, Vol. 7, Issue 4, p. 383; C.
Geiger and E. Izyumenko, “Shaping Intellectual Property Rights through Human Rights Adjudication : The Example
of the European Court of Human Rights”, Mitchell Hamline L. Rev. 2020, Vol. 46, p. 527.
14
On the notion of digital constitutionalism, see O. Pollicino, “Judicial Protection of Fundamental Rights on the
Internet, A Road towards Digital Constitutionalism?”, Bloomsbury Publishing, 2021; G. De Gregorio, “Digital
Constitutionalism in Europe, Reframing Rights and Powers in the Algorithmic Society”, Cambridge University Press,
2022. Further in the IP context, see C. Geiger and B.J. Jütte, “Designing Digital Constitutionalism: Copyright
Exceptions and Limitations as a Regulatory Framework for Media Freedom and the Right to Information Online”, in:
M. Senftleben et al. (eds.), “Cambridge Handbook of Media Law and Policy in Europe”, Cambridge University Press,
2024 (forthcoming); “Digital Constitutionalism and Copyright Reform: Securing Access to through Fundamental
Rights in the Online World”, The Digital Constitutionalist, 24 January 2022, available at http://digi-con.org/digital-
constitutionalism-and-copyright-reform-securing-access-through-fundamentalrights- in-the-online-world/; and with a
specific link to generative AI, see C. Geiger and V. Iaia, “Generative AI, Digital Constitutionalism and Copyright:
Towards a Statutory Remuneration Right grounded in Fundamental Rights”, The Digital Constitutionalist, 4 October
2023, available at https://digi-con.org/generative-ai-digital-constitutionalism-and-copyright-towards-a-statutory-
remuneration-right-grounded-in-fundamental-rights/.
15
C. Geiger and E. Izyumenko, “The Constitutionalization of Intellectual Property law in the EU and the Funke
Medien, Pelham and Spiegel Online decisions of the CJEU: Progress, but still some way to go!”, International Review
of Intellectual Property and Competition Law (IIC) 2020, Vol. 51, Issue 3, p. 282; T. Mylly, “Of Values and
Legitimacy – Discourse Analytical Insights on the Copyright Case Law of the Court of Justice of the European Union”,
Modern Law Review 2018, Vol. 81, Issue 2, p. 282.
16
C. Geiger, “The Constitutionalization of IP Law as a guaranty for competing freedoms: is the glass half empty or
half full?”, paper presented at the conference “Conflicting Freedoms in Intellectual Property Law” organized by the
EBS Law School of the University of Wiesbaden (Germany), 27 October 2023 (on file with author).
17
For further discussion of the principle of proportionality used to mitigate conflicts between fundamental rights see
J. Christoffersen, “Human Rights and Balancing: The Principle of Proportionality”, in C. Geiger (ed.), “Research
Handbook on Human Rights and Intellectual Property”, supra note 13 at p. 19; O. Fischman Afori, “Proportionality –
A New Mega Standard in European Copyright Law”, IIC 2014, Vol. 45, p. 889; C. Geiger and E. Izyumenko,
“Intellectual Property before the European Court of Human Rights”, in: C. Geiger et al. (eds.), “Intellectual Property
and the Judiciary”, supra note 13 at p. 9 sq.; C. Geiger and E. Izyumenko, “Shaping Intellectual Property Rights
through Human Rights Adjudication: The Example of the European Court of Human Rights”, supra note 13. For a
critical assessment, see T. Mylly, “Proportionality in the CJEU’s Internet Copyright Case Law: Invasive or Resilient?”,
in: U. Bernitz, X. Groussot, J. Paju, S.A. de Vries (eds.), “General Principles of EU Law and the EU Digital Order”,
Kluwer Law International, 2020, p. 257.

6
from sectorial interests. It is argued here that human rights can (even must) serve as an effective
framework to define an ethical copyright regime applicable to artificial intelligence (AI), in
particular generative AI.

In general, copyright protection can be derived from provisions protecting the right to science and
culture18, the right to freedom of (artistic) expression19, the right to protection of the moral and
material interests of the author20 and the right to (intellectual) property21. These human rights
provisions have in common an anthropocentric approach, the human author being at the center of
the protection22. Consequently, a human rights perspective mandates that the protection of creators
and human creativity is considered as the point of reference when assessing future reforms
concerning copyright and generative AI. With regards to the right to intellectual property, under
Art. 17.2 EUCF, it is to be noted that the wording of the provision as such makes no explicit
reference to the human creator.23 It could thus be argued that the scope of protection also includes

18
Arts. 27.1 of the Universal Declaration of Human Rights, 15.1a) of the International Covenant on Economic, Social
and Cultural Rights, 13 of the European Union Charter of Fundamental Rights. See on the right to culture, C. Geiger,
“Taking the Right to Culture Seriously: Time to Rethink Copyright Law”, in: C. Geiger (ed.), “Intellectual Property
and Access to Science and Culture: Convergence or Conflict?”, CEIPI/ICTSD publication series on “Global
Perspectives and Challenges for the Intellectual Property System”, 2016, Issue 3, p. 84, available at
https://www.ceipi.edu/websites/ceipi/Documents/Publications_CEIPI___ICTSD/CEIPI-ICTSD_no_3.pdf.
19
Arts. 19 of the Universal Declaration of Human Rights, 15.1 b of the International Covenant on Economic, Social
and Cultural Rights, 11 of the European Union Charter of Fundamental Rights.
20
Arts. 27.2 of the Universal Declaration of Human Rights, 15.1 c) of the International Covenant on Economic, Social
and Cultural Rights.
21
Arts. 17.2 of the European Union Charter of Fundamental Rights; 1 Protocol No. 1 to the European Convention on
Human Rights. Further on the human rights justifications for intellectual property and their roots in human rights law
see C. Geiger, “Implementing Intellectual Property Provisions in Human Rights Instruments: Towards a New Social
Contract for the Protection of Intangibles”, in: C. Geiger (ed.), “Research Handbook on Human Rights and Intellectual
Property”, Edward Elgar, 2015, p. 661 sq.
22
See in particular with regard to the international human rights framework, “UN Committee on Economic, Social
and Cultural Rights (CESCR), General Comment No. 17, “The Right of Everyone to Benefit from the Protection of
the Moral and Material Interests Resulting from any Scientific, Literary or Artistic Production of Which He or She is
the Author, Art. 15, Para. 1 (c) of the Covenant”, 12 January 2006, E/C.12/GC/17, in particular Par. 7, available at
http://www.refworld.org/docid/441543594.html; UN General Assembly, Report of the Special Rapporteur in the
field of cultural rights, Farida Shaheed, “Copyright Policy and the Right to Science and Culture”, Human Rights
Council, Twenty-eighth session, A/HRC/28/57, 24 December 2014, available at
https://digitallibrary.un.org/record/792652?ln=en, in particular Par. 90, emphasizing that a “human rights perspective
focuses attention on important themes that may be lost when copyright is treated primarily in terms of trade: the
social function and human dimension of intellectual property, the public interests at stake, the importance of
transparency and public participation in policymaking, the need to design copyright rules to genuinely benefit human
authors”(emphasis added) and that “the right to protection of authorship is the right of the human author(s) whose
creative vision gave expression to the work. Corporate right holders must not be presumed to speak for the interests
of authors” (Par. 99).
23
“Intellectual property shall be protected.”, Art. 17.2 of the European Union Charter of Fundamental Rights. See
specifically, J. Griffiths and L. McDonagh, “Fundamental Rights and European Intellectual Property Law – The Case

7
beneficiaries other than the creator himself. However, Art. 17.2 EUCF must be interpreted in the
light of other international human rights provisions which clearly centralize the protection of the
human creator. Further, the concept of the social function of intellectual property law must be
taken into account when interpreting this provision. It follows, that copyright protection should be
granted to the extent it serves the public interest24, human creativity rather than investment being
the triggering factor for protection, which leads also to an author-centric approach25. On the
contrary, the mentioned constitutional rights do not grant protection for machines or for the ones
operating the machines. Generative AI as such does not enjoy human rights protection; thus, from
a human rights perspective, the protectability of its results will be dependent on the degree of
involvement of the human creator.

As outputs produced by generative AI can be beneficial to human creators in their creative process,
the training of AI systems is essential for human beings to explore new avenues of artistic
expression that are still unknown. It should be made clear that freedom of (artistic) expression
concerns exclusively human beings considering that, at least under the current state of the law, AI
does not enjoy the mentioned constitutional right, in particular no right to free expression26. “This
implies that the interest in the flourishing of the Generative AI industry should remain instrumental
to objective of increasing human artistic freedom of expression”27.

Moreover, the outputs produced by generative AI can lead without any doubt to useful
advancements in science and arts that benefit society at large and hence can fall under the protection

of Art 17(2) of the EU Charter”, in: C. Geiger (ed.), “Constructing European Intellectual Property: Achievements and
New Perspectives”, EIPIN Series Vol. 1, Edward Elgar, 2013, p. 75; C. Geiger, “Intellectual Property shall be
protected!? Article 17 (2) of the Charter of Fundamental Rights of the European Union: A Mysterious Provision with
an Unclear Scope”, EIPR 2009, p. 113.
24
On the social function of intellectual property protection, see C. Geiger, ‘The Social Function of Intellectual Property
Rights, Or how Ethics can Influence the Shape and Use of IP Law’, in: G.B. Dinwoodie (ed.), “Intellectual Property
Law: Methods and Perspectives”, Edward Elgar, 2013, p. 153; C. Geiger, “Implementing Intellectual Property
Provisions in Human Rights Instruments: Towards a New Social Contract for the Protection of Intangibles”, supra
note 21 at p. 661; on the philosophical foundation, see D. Gervais, “Human Rights and the Philosophical Foundations
of Intellectual Property”, in: C. Geiger (ed.), Research Handbook on Human Rights and Intellectual Property, supra
note 13 at p. 89.
25
See C. Geiger, “Intellectual Property and Investment Protection: A Misleading Equation”, supra note 3 at p. 7.
26
C.R. Sunstein “Artificial Intelligence and the First Amendment”, 28 April 2023, available on SSRN:
https://ssrn.com/abstract=4431251.
27
C. Geiger and V. Iaia, “The Forgotten Creator: Towards a Statutory Remuneration Right for Machine Learning of
Generative AI”, Computer Law & Security Review, 2023 (forthcoming), available on SSRN:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4594873.

8
of the right to science and culture28. Accordingly, a right to train AI systems for artistic and
scientific purposes can be derived as a principle from the underlying human rights framework. The
human rights grounding is even stronger when the training is conducted for scientific and research
purposes as the training process can benefit additionally from the fundamental right to research-
justification29.

In order to train these generative AI systems, a large amount of copyrighted works is required. It
follows from the right to the protection of material rights of the creator, under Arts. 27.2 UDHR,
15.1 c ICESCR, and the right to (intellectual) property, under Arts. 17.2 EUCF, 1 Protocol No. 1
ECHR, that the author must be fairly renumerated in case of commercial use of his work in the
absence of justifications to do so out of competing human rights30. As argued previously, “a
conceptualization of copyright within the constitutional right to science and culture, to freedom of
expression and artistic creativity and even within a socially rooted property clause […] should lead
to the recognition of an overarching and unwritten right for creators to be remunerated for the
commercial exploitation of their work. This right should be understood as a fundamental and
binding principle of copyright law deriving from fundamental rights and from copyright’s social
function and would thus have to be recognized by legislators through the implementation of
appropriate mandatory copyright contract rules or statutory remunerations rights; it could also be

28
See in this spirit R.C. Dreyfuss, who argues that the right to “share in scientific advancement” must be re-interpreted
as a right to participate in the enterprise of scientific advancement in: “Human Rights in a Technological Age: The
Right to Participate in Science”, 15 February 2023, New York University Journal of International Law and Politics
(JILP), 2023, Vol. 55, NYU School of Law, Public Law Research Paper No. 23-35, NYU Law and Economics
Research Paper No. 23-31, available on SSRN: https://ssrn.com/abstract=4360115.
29
For a conceptualization of the fundamental right to research emerging from norms, concepts, interpretations and
understandings present in the fundamental rights of the ECHR and the CFREU, as well as international human rights
instruments, including the UCDHR, the ICCPR, and the ICESCR, see C. Geiger and B.J. Jütte, “Conceptualizing a
‘Right to Research’ and its Implications for Copyright Law, An International and European Perspective”, American
University International Law Review 2023, Vol 38, Issue 1, p. 1. See also, with regard to international law, S. Samtani,
“Developing a Human Right to Research in International Law”, (2023) PIJIP/TLS Research Paper Series No. 107,
available at: https://digitalcommons.wcl.american.edu/research/107
30
According to the case law of the ECtHR, a strong justification for the use will for example arguably be every time
the use falls within the core of freedom of expression and thus is likely to prevail in a proportionality assessment, when
the use of the work e.g. concerns political, scientific and artistic expression/ debate as well as the information of the
public on matters of public interest, and does not encroach on the very core of another competing right such as the
right to property (see detailed on the balancing criteria used by the Strasbourg Court, C. Geiger and E. Izyumenko,
“Copyright on the Human Rights Trial: Redefining the Boundaries of Exclusivity through Freedom of Expression”,
IIC 2014, Vol. 45, 316, at 322 sq). These situations do not lead to any payment, and this is why several exempted uses
in copyright law are free uses and should always remain it. However, it is to be noted that for other, less straight-
forward situations “the payment of remuneration arguably mitigates the density of the freedom of expression's conflict
with property of copyright owner and thereby increases the chances that the freedom of expression-based use would
prevail (and “fair use” will accordingly be found)” (C. Geiger and E. Izyumenko, “Towards a European ‘Fair Use’
Grounded in Freedom of Expression”, American University International Law Review 2019, Vol. 35, No. 1, at p. 58).

9
used by the judiciary as an interpretation tool of existing rules to redress unfair remuneration
situations for creators, or, in their absence, even beyond as a general principle of law”31. In contrast,
the mere amortization of economic investment in AI technology can under no circumstances be
derived from a fundamental rights perspective.

Beyond the protection of the author´s material interests, the human rights framework also confers
the protection of moral interests, under Arts. 27.2 UDHR, 15.1 c ICESCR32. More generally, moral
rights can emanate from the need to protect the authors´ dignity and personality33 – values that
underly in various fundamental right in the human rights treaties. The protection of the author´s
moral interests can be allocated to the right to privacy and personal integrity or to the so-called
“negative” aspect of the right to freedom of expression34. In the tension between copyright and
generative AI training, the moral right of the author could justify that an ability to oppose the use
of his work in certain (limited) circumstances. In fact, the weaker the fundamental rights claim to
train AI is, the stronger the moral rights claim can be. For example, training an AI to produce works
for discriminatory or racist purposes will benefit from a weaker (if any) fundamental rights
protection, but will potentially raise important moral concerns of the author of the work used for
training purposes.

31
C. Geiger, “Building an ethical framework for intellectual property in the EU: time to revise the Charter of
Fundamental Rights”, in: G. Ghidini and V. Falce (eds.), “Reforming Intellectual Property Law”, Edward Elgar, 2022,
p. 90-91. In this spirit, see also Farida Shaheed, “Copyright Policy and the Right to Science and Culture”, supra note
22, Par. 100: “Merely enacting copyright protection is insufficient to satisfy the human right to protection of authorship.
States bear a human rights obligation to ensure that copyright regulations are designed to promote creators’ ability to
earn a livelihood (…)”.
32
By contrast to the international human rights law instruments at European level an elaborated and balanced clause
for IP protection which includes moral rights is lacking, see C. Geiger and E. Izyumenko, “Designing a Freedom of
Expression-Compliant Framework for Moral Rights in the EU: Challenges and Proposals”, in: Y. Gendreau (ed.),
“Research Handbook on Intellectual Property and Moral Rights”, Edward Elgar, p. 292 sq., available on SSRN:
https://ssrn.com/abstract=3868008.
33
See e.g. P.B. Hugenholtz, “Copyright and Freedom of Expression in Europe”, supra note 1 at p. 346 (pointing to the
German case law and doctrine that recognize implied constitutional underpinnings for moral rights by situating the
interest in their protection in the German Constitution’s rights to dignity and self-fulfillment); J. Drexl, “Constitutional
Protection of Authors’ Moral Rights in the European Union – Between Privacy, Property and the Regulation of the
Economy”, in: K.S. Ziegler (ed.), “Human Rights and Private Law: Privacy as Autonomy”, Hart Publishing, 2007,
p. 159, (highlighting that, according to the continental copyright tradition of author’s rights, “the copyrighted work is
considered an emanation of the creator’s personality”. See also P. Hughes, “Painting on a Broader Canvas: The Need
for a Wider Consideration of Moral Rights Under EU Law”, EIPR 2018, Vol. 40, Issue 2, p. 95, (exploring the concept
of moral rights with reference to the fundamental right to human dignity); C. Geiger, “Droit d’auteur et droit du public
à l’information: approche de droit comparé”, supra note 1 at p. 129 sq.; “Reconceptualizing the Constitutional
Dimension of Intellectual Property – An Update”, in: P. Torremans (ed.), “Intellectual Property and Human Rights”,
supra note 13 at p. 137 (in particular Fn. 79).
34
C. Geiger and E. Izyumenko, “Designing a Freedom of Expression-Compliant Framework for Moral Rights in the
EU: Challenges and Proposals”, supra note 32.

10
In short, these general principles deriving from the human rights framework, notably the centrality
of the human creator, should inform the copyright reforms with regards to generative AI systems.

III. The protectability of AI-generated works

Before considering the policy question of whether (or not) it is desirable to protect works generated
by AI, it is interesting to analyze if de lege lata an AI-generated work can already meet the
conditions for protection.

In order to be protected, it is obvious that the work generated by an AI must meet the conditions
set forth by copyright law. Many (excellent) writings have recently been devoted to this issue so
that a detailed analysis will not be conducted here35. Let it just be recalled that according to
copyright law, the condition is twofold: first, there must be a protectable “work” and second, this
work must be “original”36. What is a “work” is generally not defined by the legislator at national,
EU or international level; however, it is commonly accepted that human intervention is necessary
for a work to be protected under copyright law. According to some authors, this obviously stems
from copyright law’s emphasis on the creator, whose creative act is the generator of the protection
granted37. In fact, the notion of originality, defined by the Court of Justice of the EU as the “author’s
own intellectual creation”38 implies creative choices39. This inevitably necessitates the ability to

35
See notably P.B. Hugenholtz and J.P. Quintais, “Copyright and Artificial Creation: Does EU Copyright Law Protect
AI-Assisted Output?”, International Review of Intellectual Property and Competition Law 2021, Vol. 52, p. 1190; D.
Gervais, “The Machine as Author”, Iowa Law Review, 2020, Vol. 105, p. 2053; J.C. Ginsburg, “People Not Machines:
Authorship and What It Means in the Berne Convention”, IIC 2018, Vol. 49, p. 131; J.C. Ginsburg and L. A. Budiardjo,
“Authors and Machines”, Berkeley Technology Law Journal 2019, Vol. 34, Issue 2, p. 343; A. Guadamuz, “Do
Androids Dream of Electric Copyright? Comparative Analysis of Originality in Artificial Intelligence Generated
Works”, in: R.M. Hilty, J.-A. Lee and K.-C. Liu (eds.), “Artificial Intelligence and Intellectual Property”, Oxford
University Press, 2021, p. 147; D. Gervais, “The Human Cause”, and G. Frosio, “Four Theories in Search of an
A(I)uthor”, in: R. Abbott (ed.), “Research Handbook of Artificial Intelligence and Intellectual Property”, supra note 4
at p. 22 and 156; M. Vivant, “Intelligence artificielle et propriété intellectuelle”, Comm. Com. Électr., 2018, Study No.
18; S. Dusollier, “Du commun de l’intelligence artificielle”, A. Bensamoun, “Libres propos sur l’existence d’un droit
de l’œuvre applicable aux créations issues de l’intelligence artificielle”, N. Binctin, “L’influence de l’intelligence
artificielle sur les mécanismes de la propriété intellectuelle”, in : J.-M. Bruguière and C. Geiger (eds.), “Penser le droit
de la pensée, Mélanges en l’honneur de Michel Vivant”, Dalloz, 2020, pp. 107, 13 and 41.
36
On this latter condition in the context of AI generated or AI-assisted works, see V. Iaia, “To Be, or Not to Be …
Original Under Copyright Law, That Is (One of) the Main Questions Concerning AI-Produced Works”, GRUR Int.
2022, Vol. 71, Issue 9, p. 793 sq.
37
See in this sense, notably, D. Gervais, P.B. Hugenholtz and J.P. Quintais, J. Ginsburg, supra note 34.
38
CJEU – Judgment of 16 Jul 2009, C-5/08 (Infopaq), Par. 37.
39
The link with the human author is even more present in the definition of originality according to French copyright
law where it is understood as “the imprint of the personality” of the author. See in this sense, M. Vivant, “Intelligence
artificielle et propriété intellectuelle”, supra note 35, underlining that according to a “personalist” copyright
conception, it is the imprint of the personality of the author that make a work protectable by copyright law, and this
personality is necessarily and intrinsically linked to a human person.

11
make conscious decisions, thus requiring a “mind”, a “heart” or a “soul” to use Kazuo Ishiguro’s
poetic imagery quoted above40. In a less poetic way, this is exactly what the U.S. Copyright Office
recalled when it held that for the work to be protectable, it must emanate from a human being 41.
The Office had already decided in a similar way when specifying that a photo taken by a monkey
cannot be registered as a work42. Reference is made here to the famous self-portrait taken by a
monkey, the “Monkey Selfie”, which had given rise to a dispute over copyright ownership of the
photo between the photographer who owned the camera used by the said monkey and the
association PETA (People for the Ethical Treatment of Animals)43. Therefore, the question: “Is a
work produced exclusively by an AI protected by copyright?” can be answered negatively based
on current understanding of copyright law44.

40
See in this sense P. Keller, “Copyright rules for AI generated visuals will determine the future of synthetic worlds”,
9 September 2022, available at https://openfuture.eu/blog/copyright-rules-for-ai-generated-visuals-will-determine-
the-future-of-synthetic-worlds/, who points out, about image-generating AIs: “On a fundamental level, they are – like
all other computers – copying machines. While there is a lot of hype and awe around the new crop of image generators
that can indeed generate stunning visual output in reaction to textual prompts fed into them, this does not mean that
they are somehow capable of independently creating works of art”. See also P. Samuelson, C.J. Sprigman, and M. Sag,
“Comments in Response to the Copyright office’s Notice of Inquiry on Artificial Intelligence and Copyright”, 30 Oct.
2023, p. 2 sq (available at https://www.regulations.gov/comment/COLC-2023-0006-8854): “An AI system can’t
produce works that reflect its own “original intellectual conception” because an AI system is incapable of having one.
AI systems do not “think” or “create” as we understand those terms in the context of human mental processes. Rather,
AI systems employ math to make predictions (…). Thus, the notion of AI being recognized as an author is a doctrinal
non-starter”.
41
See also the U.S. Copyright Office Guidelines: “Compendium of U.S. Copyright Office Practices”, 3rd ed., 2014,
updated on 28 January 2021, par. 306: “The Human Authorship Requirement: The U.S. Copyright Office will register
an original work of authorship, provided that the work was created by a human being. The copyright law only protects
“the fruits of intellectual labor” that “are founded in the creative powers of the mind”. Because copyright law is limited
to “original intellectual conceptions of the author”, the Office will refuse to register a claim if it determines that a
human being did not create the work”, available at https://www.copyright.gov/comp3/docs/compendium.pdf.
42
Ibid., para. 313.2: “Works That Lack Human Authorship: The U.S. Copyright Office will not register works
produced by nature, animals, or plants. Examples: A photograph taken by a monkey”.
43
Naruto v. Slater, 888 F.3d 418, 437 n.11 (9 th Cir. 2018). The Court did not rule on the merits, as the association’s
action on behalf of the monkey against the photographer was dismissed on procedural grounds.
44
A notable exception is the United Kingdom’s Copyrights, Designs and Patents Act (CPDA) of 1988, which allows
to protect “computer-generated works”, these works being defined according to Section 178 CPDA as works
“generated by computer in circumstances such that there is no human author of the work”. According to Art. 9(3) of
the same Act, “the author shall be taken to be the person by whom the arrangements necessary for the creation of the
work are undertaken”. This provision has been however considered by some authors as being in contradiction with
Section 1 of the same Act which specifies that copyright subsists in “original literary, dramatic, musical or artistic
works”. According to M. Blaszczyk, “Contradictions of Computer-Generated Works’ Protection”, Kluwer Copyright
Blog 6 Nov. 2023, https://copyrightblog.kluweriplaw.com/2023/11/06/contradictions-of-computer-generated-works-
protection/, the UK Copyright Act “provides for protection of authorless works, ascribing authorship to the person
making the necessary arrangements, but according to Section 1, without authorial originality, no works are protectable.
In this way, the contradictions of the statutory text, mirror the contradiction involved in the very concept of “emergent”
or “authorless” works. Without a human author, there is no expression of ideas that can be original, and thus no
copyrightable work. The concept of computer-generated works is thus logically inconsistent and incoherent with all of

12
That said, it would be wrong to stop there. As a study conducted for the European Commission has
pointed out, if a work solely created by an AI cannot be protected by copyright, works for which
the creators use the AI as a tool in the production process can be, provided of course that the
conditions of originality and the existence of creative choices are met45. With regard to copyright
principles, this is nothing new since it is accepted that an author can use a computer program or a
machine for the realization of his/her creation from the moment he/she makes creative choices.
This does not exclude a part of randomness in the result when consciously included the creation
process. The photographer presses the button, but it is the camera (and its various functions, often
automatic, such as filters) that takes the picture46. Sometimes, moreover, it is the creator’s choices
alone that will be at the origin of the protection granted. Contemporary art provides numerous
illustrations of this. When Duchamps “chooses” a commercially purchased urinal for display, it is
the choice, not the making, that constitutes the creative act at the origin of the copyright. In the
context of databases, it is the creative choices in the arrangement, selection and presentation of the
database which will trigger copyright protection.

To return to AI-generated “works”, human creative intervention can occur at different stages of
their production: Hugenholtz et al. proposes to distinguish the “conception” (design and
specifications), the “execution” (producing draft version) and “redaction” (selecting, editing,
refinement, finalization)47. The latter corresponds to the stage of finalization of the work, which
will see the AI-generated result reworked, refined, edited and modified by human intervention
(which, when it is creative and meets the conditions of copyright, can lead to copyright

copyright’s doctrinal architecture”. See also in this sense M. Blaszczyk, “Impossibility of Emergent Works’ Protection
in U.S. and EU Copyright Law”, North Carolina J.L. & Tech. 2023, Vol 25, Issue 1.
45
P.B. Hugenholtz et al., “Trends and developments in artificial intelligence – Challenges to the intellectual property
rights framework – Final report”, Publications Office of the European Union, 2020, available at
https://data.europa.eu/doi/10.2759/683128 ; For a synthetic version of the study, see P.B. Hugenholtz and J.P. Quintais,
“Copyright and Artificial Creation: Does EU Copyright Law Protect AI-Assisted Output?”, supra note 56 at p. 1190
and its comment by C. Geiger, “The Role of Human Creativity in the Copyrightability of Artificial Intelligence-
Generated Works”, JOTWELL, 29 April 2022, available at https://ip.jotwell.com/the-role-of-human-creativity-in-the-
copyrightability-of-artificial-intelligence-generated-works/.
46
For such an analogy, see C. Laser, “How A Century-Old Insight of Photography Can Inform Legal Questions of AI-
Generated Artwork”, Technology & Marketing Law Blog, 2 August 2023, available on
SSRN: https://ssrn.com/abstract=4529595.
47
P.B. Hugenholtz et al., “Trends and development in artificial intelligence: challenges to the intellectual property
rights framework”, supra note 45. See also O. Bulayenko, J.P. Quintais, D. Gervais and J. Poort, “AI Music Outputs:
Challenges to the Copyright Legal Framework”, reCreating Europe Report, February 2022, p. 109, available on SSRN:
https://ssrn.com/abstract=4072806: “Absent additional transparency measures, an accurate substantive assessment of
originality will require reverse engineering of the human interventions or contributions in the use of the AI system
leading to a certain output”.

13
protection)48. As the condition of originality, at least according to the case law of the CJEU, is very
easily met, judges often finding creativity in productions with a very low personal touch, the
exclusion of protection for AI-generated works could be circumvented quite easily as long as there
is creative human intervention in the final result49. For these reasons, several scholars have
considered that there is no need for legislative intervention at the EU or US level because the
current copyright framework already allows for protection of most AI-generated works50.

Thus, the question arising from a de lege lata perspective is howmuch human creative intervention
on the AI-generated output is necessary to trigger copyright protection. There is certainly a grey
zone here that will lead to difficult case-by-case decisions. AI technology is already frequently
used by artists in their creative process and AI-based artistic practices are likely to augment in the
future51. As an illustration, the Museum of Modern Art (MoMA) has recently acquired Refik
Anadol’s artwork “Unsupervised- Machine Hallucinations”, marking the first time the institution
has ever added piece created using mainly AI technology to its collection.52 According to digital

48
See also M. Fenwick and P. Jurcys, “Originality and the future of copyright in an age of generative AI”, Computer
Law & Security Review: The International Journal of Technology Law and Practice 2023 (forthcoming), describing
“an iterative, dynamic process of conception, prompting, generation, refining, and deployment to characterise
creativity in this context”.
49
See in this sense for the US (but with a comparison to the EU), E. Lee, “Comment of Professor Edward Lee to
Artificial Intelligence Study by The United States Copyright Office”, 30 Oct. 2023, available on ssrn:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4619118: “As the Supreme Court elaborated in Feist, the proper
test of authorship examines whether the person contributes, at least, a minimal level of creativity in the origination of
the work, which may be satisfied simply by a person’s selection or arrangement of elements in the work. The requisite
level is, as the Court recognized, “extremely low,” or the bare minimum to qualify as an author” (emphasis added).
50
For the EU: P.B. Hugenholtz et al., “Trends and development in artificial intelligence: challenges to the intellectual
property rights framework”, supra note 45; M. Fenwick and P. Jurcys, “Originality and the future of copyright in an
age of generative AI”, supra note 48, who “contend that copyright – specifically the concept of originality as articulated
in US federal law – is a sufficiently durable legal mechanism that can manage these new cultural forms, and that the
two basic requirements of modern copyright law (a tangible medium of expression and a modest degree of creativity)
remain relevant in identifying the scope of legal protection”; C. Saiz García, “Las obras creadas por sistemas de
inteligencia artificial y su protección por el Derecho de autor”, InDret 2019, Issue 1, available at https://indret.com/las-
obras-creadas-por-sistemas-de-inteligencia-artificial-y-su-proteccion-por-el-derecho-de-autor/, analyzing “the
possible protection of AI-created works under existing copyright law without the need for modification because there
is human activity behind them”; K. Militsyna, “Human Creative Contribution to AI-Based Output – One Just Can(’t)
Get Enough”, GRUR Int. 2023, Vol. 72, Issue 10, p. 939 concluding that in “many cases of using AI to create output
human creative participation remains sufficient for copyright protection eligibility”; in the US, see e.g. P. Samuelson,
C.J. Sprigman, and M. Sag, supra note 40, p. 3: “We see no need for special copyright or sui generis rules for AI”.
51
See in this sense M. Fenwick and P. Jurcys, “Originality and the future of copyright in an age of generative AI”,
supra note 48, stating that “while hybrid-networked (i.e., human – corporate – machine) creators have always created
hybrid-networked cultural forms (i.e., creations that blend human and technology-constituted elements), such hybridity
becomes increasingly visible and complex in the context of a new world of generative AI. At the very least, earlier –
and influential – models of creativity as human-driven involving creation ex nihilo become harder to sustain in a new
age of generative AI”.
52
See https://www.artforum.com/news/moma-acquires-refik-anadols-unsupervised-517497/.

14
artist, he has utilized machine intelligence as a “collaborator of human consciousness […] to unfold
unrecognized layers of our external realities.”53 In order to generate the piece, Anadol and his team
trained the AI with vast amounts of data (including copyright protected works) from digital visual
archives of the MOMA and other publicly available sources, then output a series of ever-shifting
audiovisuals that reinterpret, alter, and riff on those original works54.

The “copyrightability” of this particular work has not (yet) been discussed or questioned, however
it can be expected that such challenges will happen in the future. It is thus to be hoped that in order
to distinguish the protected from the unprotected output, the human creativity factor will be
decisive, and protection granted only when the human creative input is predominant in comparison
to the one of the machines in the end result: in short, where it can be clearly demonstrated that the
machine was strictly used as a tool in a human creative process 55, that the end result was “AI
assisted” and not “AI generated”, the burden of proofs lying on the one claiming copyright
protection.

Putting human creativity at the center of the copyrightability question does however not answer
the question from a policy perspective if this should be a desirable result 56. In fact, some authors

53
See https://refikanadol.com/works/unsupervised/.
54
The field of music, an illustration of the use of GenAI in electronic music can be seen in one of the French DJ David
Guetta´s music composition played live, in which he uses generative AI tools to create lyrics and a voice in the style
of the rapper Eminem which he then mixes over own produced electro beats (See M. Fenwick and P. Jurcys,
“Originality and the future of copyright in an age of generative AI”, supra note 48, according to which this example
“reveals something more general about creativity in a digital age” and concluding that “copyright law should remain
an important mechanism to facilitate genuine creators who are using AI systems in innovative and unique ways to push
the boundaries of their creativity”). The use of ones’ voice or image can however be very problematic from the
perspective of protection of fundamental right to personality (concerns are often raised in the context of so called “deep
fakes”, see e.g. E. F. Judge and A. M. Korharni, “Deepfakes, Counterfeits, and Personality”, Ottawa Faculty of Law
Working Paper No. 2021-21, available on SSRN: https://ssrn.com/abstract=3893890; K. Tyagi, “Deepfakes, Copyright
and Personality Rights an Inter-Disciplinary Perspective”, in: K. Mathis and A. Tor (eds.), “Law and Economics of
the Digital Transformation”, Springer, 2023, p. 191. This topic, however important, goes beyond the scope of this
article and will not be addressed.
55
On the distinction between sufficient and insufficient human creative participation, see K. Militsyna, “Human
Creative Contribution to AI-Based Output – One Just Can(’t) Get Enough”, GRUR Int. 2023, Vol. 72, Issue 10, p. 939.
See also, P. Samuelson, C.J. Sprigman, and M. Sag, supra note 40, p. 2: “Humans using AI as a tool of expression
may claim authorship if the final form of the work reflects their “original intellectual conception” in sufficient detail”.
According to these authors, even “refining a series of text prompts and choosing among different outputs should
also be recognized as a way in which a human using Generative AI could meet the authorship standard”.
56
See in this sense, P. Keller, “Copyright rules for AI generated visuals will determine the future of synthetic worlds”,
supra note 40, “the real question we should ask in response to the emergence of this new class of visual creations is
not if copyright applies to them, but rather if treating them as copyrighted works can possibly result in societal harm”.

15
have argued that it would be necessary to overcome the hurdle of human intervention to protect
AI-generated creations57. The discussion is admittedly not entirely new and recalls the “copyright
without author”58-debates held at times when copyright protection for computer programs have
been introduced59. According to others, a more cautious approach needs to be taken and the access
to copyright protection strictly limited, at least until the desirability of an extension is clearly
established60. This question is obviously most complex, and the answer might depend i.a. on the
evolution of the technology and its future uses. However, it is worth briefly recalling some of the
arguments made by each side, to the extent that an evolution of copyright law is now often required.

IV. The desirability of granting copyright protection to AI-generated works

As Professors Vivant and Bruguière have written, “de lege ferenda, the discussion is open”61.
However, the decisions to be taken are so fundamental for the future of our society that the
conversation cannot be had only with lawyers; any reform would also have to rely on independent
and serious impact assessments to establish its benefit. Too often in the past, legislative
interventions were conducted under the influence of sectorial interests, without taking into account
the potential consequences of such interventions on the innovation ecosystem or their implications
on society62. A good example is the damage made at the time by the proposed directive on

57
See in particular R. Abbott and E. Rothman, “Disrupting Creativity: Copyright Law in the Age of Generative
Artificial Intelligence”, supra note 2; C. Varytimidou, “The New A(I)rt Movement and Its Copyright Protection:
Immoral or E-Moral?”, GRUR Int. 2023, Vol. 72, Issue 4, p. 357: “Nowadays art can be created by algorithms and
continental Europe’s copyright law […] no longer seems fit for purpose, persisting as it does in searching for a human
behind each creative outcome”.
58
P. Gaudrat and M. Vivant, “Marchandisation“, in: M. Vivant (ed.), “Propriété Intellectuelle et mondialisation, La
propriété intellectuelle est-elle une marchandise ?”, Dalloz, 2004, p. 44 sq. See also J.-M. Bruguière, “Le ‘droit
d’auteur économique’, Un droit d’auteur entrepreneurial perturbateur du droit d’auteur humaniste”, in: J.-M. Bruguière
and C. Geiger (eds.), “Penser le droit de la pensée, Mélanges en l’honneur de Michel Vivant”, supra note 35 at p. 731.
59
With regard to computer generated works see already the foundational work of P. Samuelson, “Allocating
Ownership Rights in Computer-Generated Works”, University of Pittsburgh Law Rev. 1986, Vol. 47 Issue 4 p. 1186.
60
See for example D. Gervais, J.C. Ginsburg, C. Craig, supra note 2 and 35; P. Samuelson, C.J. Sprigman, and M.
Sag, supra note 40. In the context of automated journalism, see A. Trapova and P. Mezei, “Robojournalism – A
Copyright Study on the Use of Artificial Intelligence in the European News Industry”, GRUR Int. 2022, Vol. 71, Issue
7, p. 589 sq., demonstrating “that the extent to which European journalism relies on assistive and generative
technologies to produce written output does not justify, from a copyright perspective, the changing of the current
anthropocentric copyright system”; “The current copyright framework is rooted in the presence of a human author and
that should remain to be so. The absence of free and creative choices should not be artificially compensated by
considerations for potential market failures if copyright protection does not arise for robojournalism output”.
61
M. Vivant and J.-M. Bruguière, “Droit d’auteur et droits voisins”, 4th ed., Dalloz, 2019, p. 164.
62
See, in this spirit, our call for the construction of a European intellectual property law based on empirical studies,
demonstrating beforehand the potential benefits of an intervention of the European legislator, C. Geiger, “The
Construction of Intellectual Property in the European Union: Searching for Coherence”, in: C. Geiger (ed.),
“Constructing European Intellectual Property: Achievements and New Perspectives”, supra note 23.

16
computer-implemented inventions, for which the European Commission ignored the vast majority
of economic studies on the issue (including those commissioned by the Commission itself), which
had pointed out the potential negative effects in terms of innovation that broad patenting of
computer programs might cause and urged the legislator to be cautious63. The result: the
Commission went ahead, leading to the most radical rejection by the European Parliament of an
intellectual property legislation in its history; in the meantime, the legal uncertainty on the
patentability of computer programs remained, given the unclear case-law of the European Patent
Office on the issue, which was also detrimental to many economic players in Europe64.

Returning to AI, studies on the potential effects on creativity of opening up copyright protection
for AI-generated creations are currently lacking65. If AI creations were easily protected, this would
mean that countless creations will potentially be granted copyright protection, since an AI has
almost unlimited production capabilities and can generate an enormous amount of new works in
record time at low cost66. This obviously raises the question of the creative spaces remaining for
human creators and the potential very deterrent effect on future creation that such a development
would have.

Moreover, a potential change in copyright law needs to be evaluated in the light of the function
that we want copyright to fulfill67. An AI does not need to be incentivized to produce, it just
implements what it has been asked to perform, and this as often as it is asked to. It does not need a

63
See R.M. Hilty and C. Geiger, “Patenting Software? A judicial and socio-economic analysis”, IIC 2005, Vol. 36,
Issue 6, p. 615 sq. with further references.
64
C. Geiger, “Les inventions mises en œuvre par ordinateur: actualité et enjeux de l’extension contemporaine de la
brevetabilité”, in: M. Dhenne and C. Geiger (eds.), “Les inventions mises en œuvre par ordinateur: enjeux, pratiques
et perspectives”, LexisNexis, 2019, p. 1 sq.
65
On the contrary, a recent study by the European Commission highlights the potential negative effects on traditional
creative sectors of extending exclusive rights to AI-generated productions: European Commission, Directorate General
for Communication Networks, Content and technology, “Study on copyright and new technologies: copyright data
management and artificial intelligence”, Publications Office of the European Union, 2022, available at
https://data.europa.eu/doi/10.2759/570559: “The feedback received seems to indicate that an additional right in favour
of machine-generated outputs might have negative impacts on the traditional creative sectors”, p. 21.
66
See D.L. Burk, “Cheap Creativity and What It Will Do”, Georgia L. Rev., 2023, Vol. 57, p. 1669 sq.: “Like other
cost-saving industrial automation, this can be expected to displace human labor and redefine human roles in
production”.
67
See e.g. C. Geiger, “The Social Function of Intellectual Property Rights, Or how Ethics can Influence the Shape and
Use of IP law”, supra note 24, p. 153. For a reflection of AI authorship in the light copyright’s function, see also C.J.
Craig, “The AI-Copyright challenge: tech neutrality, authorship and the public interest”, in: R. Abbott (ed.), “Research
Handbook of Artificial Intelligence and Intellectual Property”, supra note 4 at p. 134, according to whom it is necessary
to resist “calls to extend copyright to cover AI-generated works on the basis that they are not equivalent to the works
of authorship that copyright seeks to encourage”.

17
break to eat, sleep, or to search for inspiration. In fact, those who emphasize the need to protect
AI-generated works, whether by copyright68 or related rights69, do not highlight the need to reward
creators but rather the objective to encourage investment in the field of AI. This well-known mantra
has become an almost Pavlovian reflex of investors in an economy now based mostly on intangible
assets: “I have invested, so I must benefit from an intellectual property right”. The issue is thus
closely related to the role of investment protection within intellectual property law. As Reto Hilty
et al. have rightly underlined, “potential protection regimes for AI – if ever required – would not
be looking at creators or inventors, but at investors”, concluding that “most AI applications lack a
theoretical justification for creating exclusive rights. If this fact is ignored, such legislation could
ultimately lead to dysfunctional effects that have negative impacts on social welfare”70.

68
See in particular R. Abbott and E. Rothman, “Disrupting Creativity: Copyright Law in the Age of Generative
Artificial Intelligence”, supra note 2 at p. 4: “Rather than acting directly on authors, copyright protection will motivate
people upstream of the creative act to use and develop AI that will result in more production and dissemination of
works”; A. Bensamoun, and J. Farchy, “Mission intelligence artificielle et culture”, Conseil Supérieur de la Propriété
Littéraire et Artistique, January 2020, p. 30, available at
https://www.culture.gouv.fr/content/download/281441/file/CSPLA-Rapport-complet-IA-
Culture_janv2020.pdf?inLanguage=fre-FR.
69
M. Senftleben and L. Buijtelaar, “Robot Creativity: An Incentive-Based Neighboring Rights Approach”, EIPR 2020,
Vol. 42, p. 797; A. Ramalho, “Will Robots Rule the (Artistic) World? A Proposed Model for the Legal Status of
Creations by Artificial Intelligence Systems”, Journal of Internet Law 2017, Vol. 21, Issue 1, p. 12; N. Gervassis and
A. Trapova, “UKIPO’s public consultation on AI and IP – computer-generated works (1st part)”, 14 March 2022,
available at http://copyrightblog.kluweriplaw.com/2022/03/14/ukipos-public-consultation-on-ai-and-ip-computer-
generated-works-part-1/; A. Shtefan, “Creations of artificial intelligence, In search of the legal protection regime”,
JIPITEC 2023, Vol. 14, Issue 1, p. 94; C. Varytimidou, “The New A(I)rt Movement and Its Copyright Protection:
Immoral or E-Moral?”, supra note 57 at p. 362, proposing a sui generis economic right that “serves as a countermeasure
and aims to provide humans with the incentive to continue to invest in AI even though humans will not be deemed to
be authors of AGA”; E. Bonadio, N. Lucchi and G. Mazziotti, “Will Technology-Aided Creativity Force Us to Rethink
Copyright’s Fundamentals? Highlights from the Platform Economy and Artificial Intelligence”, IIC 2022, Vol. 53,
p. 1174 (at 1196), arguing for some kind of sui generis right with reference to the precedent of the sui generis right for
producers of non-original databases granted by the 1996 EU Database Directive; A. Papadopoulou, “Creativity in
crisis: are the creations of artificial intelligence worth protecting?”, JIPITEC 2021, Vol. 12, p. 408: “Legislating a sui
generis right in order to boost innovation, protect competition and maintain a healthy market for intellectual creations
is suggested as the best option”.
70
R.M. Hilty, J. Hoffmann and S. Scheuerer, “Intellectual Property Justification for Artificial Intelligence”, in: R.
Hilty, J.-A. Lee and K.-C. Liu (eds.), “Artificial Intelligence and Intellectual Property”, Oxford University Press, 2021,
p. 50 sq. In this sense see also A. Ramalho, “Will Robots Rule the (Artistic) World? A Proposed Model for the Legal
Status of Creations by Artificial Intelligence Systems”, supra note 69 at p. 12, stating that “justifications for granting
copyright protection do not fit Ais creations, and privatization through the grant of (exclusive) rights should not be
readily chosen without further thought”, and with regard to the creation of a new related right, see M. Duque Lizarralde
and C. Meinecke, “Authorless AI-assisted productions: Recent developments impacting their protection in the
European Union”, JIPITEC 2023, Vol. 14, Issue 1, p. 93: “Up to date there is neither economic nor theoretical
justification (e.g., deontological or naturalistic), supporting that this related right would incentivise the creation of
authorless AI-assisted productions, instead of producing saturation in the market”. See also P. Samuelson, C.J.
Sprigman, and M. Sag, supra note 40, p. 3

18
The quasi-systematic push for IP protection is (unfortunately) not a new issue either. As it has been
often underlined, intellectual property rights tend to progressively shift to investment-protection
mechanisms71. The multiplication of neighboring and sui generis rights is surely a consequence of
this trend, as is the progressive extension of the scope of protection72. However, this multiplication
of intellectual property rights, sometimes on the same object, has not remained without
consequences and has produced a “legal hamburger in which several layers of rights overlap”73.
These layers consist of legal hurdles to overcome in the form of multiple authorizations to request,
all resulting in potential limitations for those who want to create (not to mention the uncertainty
that often surrounds the ownership of rights). To take only one example, the sui generis protection
for database producers: subsequent evaluations by the European Commission have failed to
determine whether the introduction of this right has increased the production of databases within
the EU74. Knowing that once an intellectual property right is in place it is almost impossible to
repeal it75, past experiences in the EU should lead to a cautious approach before protecting AI-
generated productions by copyright and granting rights to investors76. On the contrary, the trend

71
See also C. Geiger, “Intellectual Property and Investment Protection: A Misleading Equation”, supra note 3 at p. 7
sq.
72
For a critique of this trend, see P.B. Hugenholtz, “Neighboring Rights are Obsolete”, IIC 2019, Vol. 50, p. 1006; C.
Sganga, “The Many Metamorphoses of Related Rights in EU Copyright Law: Unintended Consequences or Inevitable
Developments?”, GRUR Int. 2021, Vol. 70, Issue 9, p. 821; C. Geiger, “Intellectual Property and Investment
Protection: A Misleading Equation”, supra note 3.
73
P. Gaudrat and M. Vivant, “Marchandisation”, in: M. Vivant (eds.), “Propriété Intellectuelle et mondialisation, La
propriété intellectuelle est-elle une marchandise ?”, supra note 57 at p. 42.
74
In this sense M. van Eechoud, “Please share nicely – From Database directive to Data (governance) acts”, Kluwer
Copyright Blog, 18 August 2021, available at http://copyrightblog.kluweriplaw.com/2021/08/18/please-share-nicely-
from-database-directive-to-data-governance-acts/, according to whom the “highly critical 2005 evaluation report of
the Database directive already signaled that the economic impact of the sui generis right was unproven, and that it
comes perilously close to an undesirable property right in data as such. The database industry (its European powerhouse
being the U.K.) then did not favour a repeal of the sui generis right and the Commission identified various other
drawbacks, so no action followed. The 2018 review of the Database directive and accompanying public consultation
perpetuated this stasis. It too concludes there is no proven economic benefit”.
75
See, on the topic, M. Husovec, “The Fundamental Right to Property and the Protection of Investment: How Difficult
Is It to Repeal New Intellectual Property Rights?”, in: C. Geiger (ed.), “Research Handbook of Intellectual Property
and Investment Law”, Edward Elgar, 2020, p. 385.
76
In this sense, see also the study of the European Commission, “Study on copyright and new technologies: copyright
data management and artificial intelligence”, available at https://op.europa.eu/en/publication-detail/-
/publication/cc293085-a4da-11ec-83e1-01aa75ed71a1/language-en, p. 21: “The research, interviews, and surveys
conducted within the study indicate firstly that no incentive for the use of AI tools in the creative process in the form
of additional exclusive rights appears necessary. The already broad deployment of AI tools in the creative context
confirms this. The study concludes that a new related right for AI-generated outputs is not desirable”; O. Bulayenko,
J.P. Quintais, D. Gervais and J. Poort, “AI Music Outputs: Challenges to the Copyright Legal Framework”, supra note
47 at p. 113 sq: “There is no clear case for a legislative action at the level of substantive rules in the EU copyright
acquis in the short term as regards AI outputs. Existing proposals for new rights and forms of protection for AI outputs
generally lack clear and convincing theoretical and economic justification […] Considering this, we recommend that
no new protection regimes for AI outputs are introduced absent clear and compelling evidence that justifies a change

19
toward overprotection of intellectual property rights should prompt prior reflection on the
justifications for granting them77. If IP rights aim at encouraging creation and protecting creators,
creativity – not investment – should be the triggering factor for protection78. On the other hand, if
the goal is to create mechanisms for investment protection, then the system cannot continue to lie
on the generous protection justified by the protection of the human creator and would need
reform79.

To sum up: taking the binding human rights framework as guidance for copyright protection leads
to an overall cautious approach when protecting AI-generated works by copyright and/ or granting
rights to investors. AI-generated output “as such” lacks the traditional protectability that human-
created works enjoy. An analogy can be drawn to Article 52 of the European Patent Convention,
which specifies certain items that are not regarded as inventions. One of these exclusions concerns
computer programs. In a similar vein, it could be decided that AI-generated content, in its raw,
unaltered form, may not be considered as a copyrighted work. This exclusion would not prevent
the combination of AI and creative human input, put set a clear signal from the legislator in
particular to the judiciary in order to avoid that the unwritten condition of human authorship be
overruled in the future. It also would induce, in the spirit of the exclusion of patentability of
computer programs “as such”, a higher level of argumentation to show that the human creative
input was determinant and predominant in the final result. Only when AI would serve as a tool of
human creativity, it could result in a copyrighted work. In such cases, the unique blend of AI's
capabilities and human ingenuity can produce content that is eligible for copyright protection.

to the status quo”; J. Drexl, R.M. Hilty, et al. “Artificial Intelligence and Intellectual Property Law, Position Statement
of the Max Planck Institute for Innovation and Competition of 9 April 2021 on the Current Debate”, Max Planck
Institute for Innovation and Competition Research Paper No. 21-10, available on SSRN:
https://ssrn.com/abstract=3822924, p. 5: “Introducing a new protection regime (e.g. a new related right) for AI-
generate output is not justified according to the current state of Knowledge”.
77
See R.M. Hilty, “The Expansion of Copyright Law and its Social Justification”, in: C. Heath and K.-C. Liu (eds.),
“Copyright Law and the Information Society in Asia”, Hart Publishing, 2007, p. 1. But there is a vast body of literature
criticizing the expansion of IP rules beyond its traditional borders. See e.g. R.C. Dreyfuss, D.L. Zimmerman and H.
First (eds.), “Expanding the Boundaries of Intellectual Property”, Oxford University Press, 2001.
78
For a recent reflection in this direction see C. Geiger, “Intellectual Property and Investment Protection: A Misleading
Equation”, supra note 3.
79
For example, protecting a work 70 years after the death of the author implies that the event giving rise to the
protection is a creation that emanates from an author (human, because the machine does not die). However, if one sets
up a “copyright without author”, this cannot remain without consequence on the duration of the right. If it is a question
of making an investment profitable, the duration can even be very short.

20
However, when determining the scope of protectability, it will be necessary to examine very closely
whether these extensions of exclusive control provide sufficient spaces of freedom to guarantee
follow-on creations and future creativity. If one wants to incentivize the development of AI-
generated creations, it would probably be more efficient further widen the scope of certain
exceptions to copyright and database right, such as the exceptions allowing text and data mining.

This leads us to the third crucial question with regard to AI-created works: their legality with regard
to copyright law. In short, after discussing the protectability of the output of the machine, the
question of the legality of the input necessary for the machine to be able to generate new works
needs to be addressed.

V. The legality of AI-generated content trained on copyright protected works

AI systems do not produce from scratch. They have to be trained through the process of machine
learning with existing data and works80. As Martin Senftleben sharply describes it, “generative AI
systems are only capable of mimicking human creativity because human works have been used as
training material. On the basis of existing literary and artistic creations that serve as input data,
machine-learning algorithms are able to recognize patterns and similarities. Following this
deductive method, a generative AI system learns how to produce novel literary and artistic output
by imitating the style of human works. The machine-learning algorithm enables the generative AI
system to generate literary and artistic content on its own – based on the computational analysis of
human works that served as training material”81. In short, to produce a work “in the style of
Picasso”, the machine will need to be trained on a high number of existing works of Picasso. The
better (and more comprehensive) the training will be, the better the results obtained82. This of
course begs the question of how this training can be evaluated from a copyright perspective, in

80
On the technical aspects of machine learning see J. Drexl, R.M. Hilty et al., “Technical Aspects of Artificial
Intelligence: An Understanding from an Intellectual Property Law Perspective”, Max Planck Institute for Innovation
and Competition Research Paper No. 19-13 (2019), available on SSRN: https://ssrn.com/abstract=3465577.
81
M. Senftleben, “Generative AI and Author Remuneration”, IIC 2023, Vol. 54 (forthcoming), available at
https://link.springer.com/article/10.1007/s40319-023-01399-4 and on SSRN:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4478370.
82
See on this issue in particular A. Levendowski, “How Copyright Law Can Fix Artificial Intelligence’s Implicit Bias
Problem”, Wash. L. Rev. 2018, Vol. 93, p. 579. In this respect, the author underlines that “copyright law causes friction
that limits access to training data and restricts who can use certain data. This friction is a significant contributor to
biased AI” (at p. 589).

21
particular when the algorithm is “fed” with copyright protected works83. The answer is not easy,
as until today no legislation has been passed to deal exactly with this situation. Therefore, one is
left with the general principles of copyright law and with the existing rules that could apply to the
situation.

Before taking a closer look at the European framework, it is worth having a look at how other
jurisdictions might handle the tension between copyright and A.I. learning. In the US, it is generally
considered84 that text and data mining (TDM) which is at the core of the machine learning process
falls under the fair use exception, based on the application by analogy of Authors Guild v. Google85
and Authors Guild v. HathiTrust86. It has to be mentioned however that these decisions were issued
before generative A.I. systems had bloomed. It is not excluded that the wide range of possible
applications on a scalable level could lead US judges to another conclusion when applying the fair-
use test under U.S. copyright law. The fourth factor87 of the fair-use analysis, which focuses on the
effects on the market, seems to be the problematic point in the AI context, as direct competition
that AI-generated works can have with those created by physical persons may lead to chilling
effects on human creativity and the earnings of creators88. In the case of Authors Guild v. Google,
Plaintiffs made three arguments why Google´s service should not be considered fair use. First, the
plaintiffs contended that Google’s service to provide digital copies of entire books, allowing users
through the snippet function to read parts of the book, provides a substitute for Plaintiffs’ work,
which would negate the possibility of a fair use finding. Second, Authors Guild alleged Google to
infringe their derivative rights in search functions, depriving Plaintiffs of revenues or other benefits

83
See on the issue A. Guadamuz, “A Scanner Darkly: Copyright Liability and Exceptions in Artificial Intelligence
Inputs and Outputs”, 26 February 2023, available on SSRN: https://ssrn.com/abstract=4371204; N. Lucchi, “ChatGPT:
A Case Study on Copyright Challenges for Generative Artificial Intelligence Systems”, European Journal of Risk
Regulation 2023, p. 1, available at https://doi.org/10.1017/err.2023.59.
84
P. Samuelson, “Text and Data Mining of In-Copyright Works: Is It Legal?”, Communications of the ACM 2021,
Vol. 64, Issue 1, p. 20; M.A. Lemley and B. Casey, “Fair Learning”, Texas Law Review 2021, Vol. 99, Issue 4, p. 743;
M. Carroll, “Copyright and the Progress of Science: Why Text and Data Mining Is Lawful”, U.C. Davis L. Rev. 2019,
Vol. 53, p. 893; C.J. Craig, “The AI-Copyright Challenge: Tech-Neutrality, Authorship, and the Public Interest”, supra
note 67 at p. 152.
85
U.S. Court of Appeals for the Second Circuit, Authors Guild, Inc. v. Google, decision of 16 October 2015, No. 804
F.3d 202.
86
U.S. Court of Appeals for the Second Circuit, Authors Guild, Inc. v. HathiTrust, decision of 10 June 2014, No. 755
F.3d 87.
87
The four factors are listed in Section 107 of the 1976 U.S. Copyright Act.
88
See more detailed on the issue of generative AI and fair use in the US: M. Sag, “Copyright Safety for Generative
AI”, Houston Law Review 2023, Vol. 62, Issue 2 (forthcoming); see also the overview from the Congressional
Research Service, see C. Zirpoli, “Generative Artificial Intelligence and Copyright Law”, 2023, available at
https://crsreports.congress.gov/product/pdf/LSB/LSB10922.

22
they would gain from licensed search markets. Third, Google’s storage of digital copies exposes
Plaintiffs to the risk that hackers will make their books freely (or cheaply) available on the Internet,
destroying the value of their copyrights. However, the Court rejected all of these arguments, main
reason for which several scholars have argued that the use of copyrighted works to train the AI is
likely to be considered fair-use under US law.89 Nevertheless, scholars are closely monitoring
several lawsuits against AI system producers in the US, as rightholders are claiming that these uses
are unfair and therefore are not covered by the fair use exception of US Copyright law.90 There are
initial indications that the rightholder lawsuits will not be successful. For example, a US District
Court in a recent decision Andersen v. Stability AI Ltd dismissed most of the claims that images
generated by the AI systems based on text prompts violated copyrights of the plaintiffs.91 The
various lawsuits taken in the U.S. will soon bring first elements on the legality of the use of works
for ML purposes. It will be interesting to see if the upcoming U.S. caselaw will be influenced by
the results in the EU or if the U.S. will focus on a completely different approach.

An illustrative instance of a maximalist approach can be seen in the recent legislative proposal
unveiled in France on September 12, 202392. This proposal suggests placing the machine learning
process under the exclusive control of rightholders whose works are used in the ML process.
Furthermore, it advocates for the attribution of authorship of the AI Generated output to all the

89
M.A. Lemley, ‘How Generative AI Turns Copyright Law on its Head’, Colum. Sci. & Tech. L. Rev. 2024,
(forthcoming); M. Sag, “The New Legal Landscape for Text Mining and Machine Learning”, Journal of the Copyright
Society of the USA 2019, Vol. 66, p. 291; M. Sag, “Copyright Safety for Generative AI”, supra note 88; P. Henderson
et al., “Foundation Models and Fair Use”, 2023, p. 5, available at https://arxiv.org/abs/2303.15715; M.A. Lemley and
B. Casey, “Fair Learning”, supra note 84 at p. 748; B.L.W. Sobel, “Artificial Intelligence’s Fair Use Crisis”, Colum.
J.L. & Arts 2017, Vol. 41, p. 45, 96.
90
See e.g. in the US: Doe 1 et al v. GitHub et al., Case No. 4:2022cv06823 (N.D. Cal.); Andersen et al v. Stability AI
et al., Case No. 3:23-cv-00201 (N.D. Cal.); Getty Images v. Stability AI, Case No. 1:2023cv00135 (D. Del.); Tremblay
P. and Awad M. v. OpenAI INC. et al., Case No. 3:23-cv-03223 (N.D. Cal.), on this latter class action, see G. Campus,
“Generative AI: the US Copyright class action against OpenAI”, Kluwer copyright blog, 14 August 2023, and in UK:
Getty Images v Stability AI, Case IL-2023-000007. More generally on these lawsuits see G. Frosio, “Generative AI in
Court”, in: N. Koutras and N. Selvadurai (eds.), “Recreating Creativity, Reinventing Inventiveness – International
Perspectives on AI and IP Governance”, Routledge, 2023 (forthcoming).
91
United States District Court of Northern California, Case No. 23-cv-00201-WHO, 30 October 2023, available at
https://drive.google.com/file/d/19oLqGeezldu1de366DMGrVYJsKXDgJS2/view. According to US District Court
Judge William Orrick, the claims of copyright infringements were not enough substantiated: “Finding that the
Complaint is defective in numerous respects, I largely GRANT defendants’ motions to dismiss and defer the special
motion to strike. Plaintiffs are given leave to amend to provide clarity regarding their theories of how each defendant
separately violated their copyrights, removed or altered their copyright management information, or violated their
rights of publicity and plausible facts in support”.
92
French National Assembly, Draft Law of the 12 September 2023, No. 1630, available at https://www.assemblee-
nationale.fr/dyn/16/textes/l16b1630_proposition-loi.pdf. For a critical analysis see C. Geiger and V. Iaia, ‘The
Forgotten Creator: Towards a Statutory Remuneration Right for Machine Learning of Generative AI‘, supra note 27.

23
authors which works have been used in the machine learning process. Additionally, it mandates
labeling the resulting output as “AI-generated work” and listing the names of all authors whose
works were utilized in the training process. Implementing such a measure would necessitate
acknowledging a substantial number—potentially thousands or even millions—of authors, given
the vast datasets that Generative AI commonly trains on. On top of being impracticable, this all-
encompassing solution would likely have very detrimental consequences for the advancement of
AI systems. It surely would lead in making any jurisdiction adopting such solution very unattractive
for AI innovators in particularly start-ups who do not have the capacity to clear all the rights for
the works used in the training process.

An AI-guideline proposal by the Japanese government from early June 2023 is heading to the
complete opposite direction. The Japanese government released some of the world’s first legal
guidelines around generative artificial intelligence imagery.93 Although not implementing a new
statutory regulation, the guidelines affirmed that machine learning engineers are allowed to use any
data they can find, regardless of their copyright protection.94 Japan had already implemented one
of the most liberal laws in 2018, allowing the free use of copyrighted works for training machine
learning models as long as the purpose “is not for enjoying or causing another person to enjoy the
ideas or emotions expressed in such work.”95 This new guideline is based on a much broader
understanding of this law, allowing to use any data “regardless of whether it is for non-profit or
commercial purposes, whether it is an act other than reproduction, or whether it is content obtained
from illegal sites or otherwise.”96

93
See the presentation by the Agency for Cultural Affairs (in Japanese), available at
https://www.bunka.go.jp/seisaku/chosakuken/pdf/93903601_01.pdf.
94
“In principle, analysis of information for the purpose of AI development is possible without permission, according
to the right limitation provision.”, see the presentation by the Agency for Cultural Affairs, supra note 93 at p. 61
(translated by the author).
95
Article 30-4 of the Japanese Copyright Act, translation by T. Ueno, “The Flexible Copyright Exception for ‘Non-
Enjoyment’ Purposes – Recent Amendment in Japan and Its Implication”, GRUR Int. 2021, Vol. 70, Issue 2, p. 148;
For a closer look on the TDM exceptions in Japanese Copyright Law see also A. Dermawan, “Text and data mining
exceptions in the development of generative AI models: What the EU member states could learn from the Japanese
“non enjoyment” purposes?”, The Journal of Intellectual Property 2023, Vol. 26, Issue 3, p. 1 sq.
96
See statement by the Minister of Education, Culture, Sports, Science and Technology Nagaoka, available at
https://go2senkyo.com/seijika/122181/posts/685617; For the broad media response, see for example
https://technomancers.ai/japan-goes-all-in-copyright-I-apply-to-ai-training/; https://restofworld.org/2023/japans-new-
ai-rules-favor-copycats-over-artists/; https://www.reuters.com/technology/japan-leaning-toward-softer-ai-rules-than-
eu-source-2023-07-03/; https://www.natlawreview.com/article/japanese-government-identified-issues-related-ai-and-
copyrights#:~:text=Do%20Inputs%20to%20AI%20Infringe,not%20unreasonably%20harm%20creators’%20interest
s.

24
In the European Union, when having a look on the current legal provisions, since AI systems learn
from datasets using the technique of Text and Data Mining (TDM), the most obvious provisions to
be scrutinized in this respect are the newly introduced limitations and exceptions for Text and Data
Mining-purposes in the directive of 17 April 2019 on Copyright in the Digital Single Market
(CDSM-Directive). Article 3 introduces an exception for text and data mining for scientific
research which solely benefits research organisations and cultural heritage institutions while
Article 4, introduced later in the elaboration process of the directive, is not restricted to specific
institutions, and therefore could be relevant in the context of AI as these systems are usually
operated by private commercial companies not covered by article 3. According to Article 4(1),
“Member States shall provide for an exception of limitation (…) for reproductions and extractions
of lawfully accessible works and other subject matter for the purposes of text and data mining”.
Interestingly, according to Art. 4(2), “reproductions and extractions made pursuant to paragraph 1
may be retained for as long as is necessary for the purposes of text and data mining”, which could
be of relevance to solve any possible question of storage of protected works by the AI in the
learning process. However, the usefulness of this provision might be rather limited by the third
paragraph of Article 4, which conditions the application of the exception to the fact that the use of
works and other subject matters “has not been expressly reserved by their rightholders in an
appropriate manner, such as machine-readable means in the case of content made publicly available
online”. In short, rightholders can “opt out” of the exception, which potentially can make the
provision rather ineffective if there is a high number of rightholders that are doing so97. A lot has
been written on the lack of ambition of the provisions on text and data mining, and it is not the
place here to discuss them at length98. These exceptions faced strong criticism for insufficiently

97
For example, in France, the SACEM (Société des auteurs, compositeurs et éditeurs de musique), the major collective
management organization for music, announced that it will exercise the opt-out on behalf of all its 200.000 members,
making it probable that other collective management organizations in France and other European countries will follow
the same path. See “Pour une intelligence artificielle vertueuse, transparente, et équitable: la Sacem exerce son droit
d’opt-out”, 12 October 2023, available at https://createurs-editeurs.sacem.fr/actualites-agenda/actualites/la-sacem-et-
vous/pour-une-intelligence-artificielle-vertueuse-transparente-et-equitable-la-sacem-exerce-son-droit.
98
See in this regard C. Geiger, G. Frosio and O. Bulayenko, “The Exception for Text and Data Mining (TDM) in the
Proposed Directive on Copyright in the Digital Single Market – Legal Aspects”, Study for the Directorate-General for
Internal Policies of the Union, Department of Citizens’ Rights and Constitutional Affairs, European Parliament,
February 2018, available on SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3160586. For a critical
evaluation of the Directive proposal, see also C. Geiger, G. Frosio and O. Bulayenko, “Text and Data Mining in the
Proposed Copyright Reform: Making the EU Ready for an Age of Big Data?”, IIC 2018, Vol. 49, Issue. 7, p. 814, and
from the same authors: “The EU Commission’s Proposal to Reform Copyright Limitations: A Good but Far Too Timid
Step in the Right Direction”, EIPR 2018, Vol. 40, Issue 1, p. 4; European Copyright Society, “General Opinion on the
EU Copyright Reform Package”, 24 January 2017, p. 5; R.M. Hilty and H. Richter, in: R.M. Hilty and V. Moscon

25
taking into account the importance of TDM in many sectors, in particular with regard to the
development of AI activities in the EU99, while other regions of the world have more flexible
approaches in their system of existing limitations100.

In any case, what needs to be emphasized here is that the TDM exceptions were not designed to
cover machine learning by generative AI systems, meaning AI systems that can produce new works

(eds.), “Modernisation of the EU Copyright Rules, Position Statement”, MPI for Innovation and Competition Research
Paper No. 17-12, p. 25 et sq.
99
See C. Geiger, “The Missing Goal-Scorers in the Artificial Intelligence Team: Of Big Data, the Fundamental Right
to Research and the failed Text and Data Mining Limitations in the CSDM Directive”, in: M. Senftleben, J. Poort, M.
van Eechoud, S. van Gompel, N. Helberger (eds.), “Intellectual Property and Sports, Essays in Honour of P. Bernt
Hugenholtz”, Kluwer Law International, 2021, p. 383 sq.; C. Geiger, G. Frosio and O. Bulayenko, “Crafting a Text
and Data Mining Exception for Machine Learning and Big Data in the Digital Single Market”, in: X. Seuba, C. Geiger
and J. Pénin (eds.), “Intellectual Property and Digital Trade in the Age of Artificial Intelligence and Big Data”, 2018,
Issue 5, p. 95. For a (critical) analysis see also P.B. Hugenholtz, “The New Copyright Directive: Text and Data Mining
(Articles 3 and 4)”, Kluwer Copyright Blog, 24 July 2019, available at
http://copyrightblog.kluweriplaw.com/2019/07/24/the-new-copyright-directive-text-and-data-mining-articles-3-and-
4/; C. Geiger, G. Frosio and O. Bulayenko, “Text and Data Mining: Articles 3 and 4 of the Directive 2019/790/EU”,
in: C. Saiz Garcia and R. Evangelio Llorca (eds.), “Propiedad intelectual y mercado único digital europeo”, Tirant lo
Blanch, 2019, p. 27; M. Kop, “The Right to Process Data for Machine Learning Purposes in the EU”, Harvard Journal
of Law & Technology 2021, Vol. 34, Issue 1, p. 7; European Copyright Society, “Comment of the European Copyright
Society Addressing Selected Aspects of the Implementation of Articles 3 to 7 of Directive (EU) 2019/790 on Copyright
in the Digital Single Market”, 3 May 2022, available at
https://europeancopyrightsocietydotorg.files.wordpress.com/2022/05/ecs_exceptions_final-3.pdf.; M. Senftleben,
“Compliance of National TDM Rules with International Copyright Law – An Overrated Nonissue?”, IIC 2022, Vol.
53, Issue. 10, p. 1477 sq.; T. Margoni and M. Kretschmer, “A Deeper Look into the EU Text and Data Mining
Exceptions: Harmonisation, Data Ownership, and the Future of Technology”, GRUR Int. 2022, Vol. 71, Issue 8, p. 685;
R. Ducato and A. Strowel, “Limitations to text and Data Mining and Consumer Empowerment: Making the Case for
a Right to ‘Machine Legibility’”, IIC 2019, Vol. 50, Issue 6, p. 649 sq. With a similar conclusion for South America,
see M. Jackson Bertón, “Text and Data Mining Exception in South America: A Way to Foster AI Development in the
Region”, GRUR Int. 2021, Vol. 70, Issue 12, p. 1145.
100
See J. Drexl, R.M. Hilty, et al., “Artificial Intelligence and Intellectual Property Law, supra note 76: “The system
of copyright exceptions and limitations, as harmonized under the InfoSoc Directive and the Digital Single Market
Directive, is not flexible enough to enable the use of IP-protected subject-matter for the purpose of developing AI
systems”. See for example for the US and the potential legality of machine learning under the fair use doctrine M. Sag,
“The New Legal Landscape for Text Mining and Machine Learning”, supra note 88; M. Carroll, “Copyright and the
Progress of Science: Why Text and Data Mining Is Lawful”, supra note 84; P. Samuelson, “Text and Data Mining of
In-Copyright Works: Is It Legal?”, supra note 84 at p. 20 sq; M. A. Lemley and B. Casey, “Fair learning”, supra note
84 at p. 743, stating however it is not clear that the use of databases (which include copyrighted works) in machine
learning processes that generate new outputs “will be treated as a fair use under current law. Fair use doctrine in the
last quarter century has focused on the transformation of the copyrighted work. Ais aren’t transforming the databases
they train on; they are using the entire database, and for a commercial purpose at that. Courts may view that as a kind
of free riding they should prohibit”. Thus, even if TDM could qualify as a “fair use” in the past, it remains uncertain
if the use of copyrighted works via machine learning in generative AI processes will be allowed under the fair use
doctrine (see also more detailed on the issue of generative AI and fair use in the US: M. Sag, “Copyright Safety for
Generative AI”, supra note 88). For Japan, see T. Ueno, “The Flexible Copyright Exception for ‘Non-Enjoyment’
Purposes ‒ Recent Amendment in Japan and Its Implication”, supra note 95 at p. 145. However, the exception
introduced in Japanese Law in 2019 seemed also not have been taking into account generative AI training situations.

26
based on the learning from other existing works101. This emerges very clearly from the preparatory
works of the Directive, where the issue of generative AI and TDM exceptions is absent and has
obviously been overlooked. It is thus rather surprising that the Commission considers that the
existing legislative framework is perfectly fit for the purpose and that “creation of art works by AI
does not deserve a specific legislative intervention”, since the TDM exceptions with their
possibility of opt-out apply, “providing balance between the protection of rightholders including
artists and the facilitation of TDM, including by AI developers”102. The recent compromise
amendments by the European Parliament to introduce transparency obligations in the proposed AI
Act103 and the vibrant public debate104 generated by the use of existing works by AI systems for
the purpose of generating new ones seem to indicate that the discussion is not over and that these
transparency obligations might be the first step of another legislative intervention105. As we have
seen, in various jurisdictions, lawsuits have been introduced to challenge the legality of the use of
existing works to train AI systems and the first decisions are expected soon106.

101
See in this sense also J. Nordemann and J. Pukas, “Copyright exceptions for AI training data—will there be an
international level playing field?”, Journal of Intellectual Property Law & Practice 2022, Vol. 17, Issue. 12, p. 974:
“One thing is certain: there is no provision in the Directive that expressly deals with the training of AI and the
copyright-related aspects”.
102
Answer given on 31 March 2023 by Commissioner Breton on Behalf of the European Commission (Parliamentary
written question for written answer, E-000479/2023, posed on 15 February 2023 “How does the Commission plan to
regulate this use of AI, which harms artists and rights holders?”), available at
https://www.europarl.europa.eu/doceo/document/E-9-2023-000479-ASW_EN.html.
103
European Commission, 21 April 2021, Proposal for a Regulation of the European Parliament and of the Council
Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act) and Amending Certain, Union
Legislative Acts, COM(2021) 206 final.
104
The debate follows after concerns have been voiced by renown authors and collective management organisations,
which benefitted from large media coverage.
105
European Parliament, 16 May 2023, Draft Compromise Amendments on the Draft Report – Proposal for a
Regulation of the European Parliament and of the Council on Harmonised Rules on Artificial Intelligence (Artificial
Intelligence Act) and Amending Certain Union Legislative Acts, Document KMB/DA/AS, Version 1.0. The European
Parliament proposes the introduction of a new Article 28b(4) establishing that “providers of foundation models used
in AI systems specifically intended to generate, with varying levels of autonomy, content such as complex text, images,
audio, or video (“generative AI”) shall: I “without prejudice to national or Union legislation on copyright, document
and make publicly available a sufficiently detailed summary of the use of training data protected under copyright law”
(emphasis added). EU lawmakers have started the negotiations to finalise the new legislation. Trilogue meetings took
place in June, July, September and October 2023, negotiations are on-going, see
https://www.europarl.europa.eu/legislative-train/theme-a-europe-fit-for-the-digital-age/file-regulation-on-artificial-
intelligence. For a comment of the copyright aspects envisaged in the “AI Act”, see J.P. Quintais, “Generative AI,
Copyright and the AI Act”, 9 May 2023, available at https://copyrightblog.kluweriplaw.com/2023/05/09/generative-
ai-copyright-and-the-ai-act/.
106
For an overview over the ongoing cases see supra note 90.

27
Therefore, it is necessary to reflect on possible solutions in this context, as it seems evident that the
existing legal framework is not easily adaptable to deal with such an important societal question as
the use of existing protected works by AI systems. Fundamental philosophical questions are at
stake, in particular what should be the place in the future of creators in a world where works can
be created quickly and at rather low costs by AI systems. As the AI is trained with already existing
works, how to then incentivize break-through creativity, news trends, new genres, disruptive art
that break with the existing?

VI. Towards a new limitation-based remuneration right to the benefit of creators for
machine learning of generative AI

It is surely not the place here to propose a final “ready- to-go” solution to the issue. A future
regulation of AI, as we have already stated above, needs to build on a societal consensus which
requires a pluridisciplinary assessment to reach informed consent. Some preliminary thoughts can
however already be shared. First, applying Article 4 CDSM and its “opt out”-mechanism to
generative AI is not a satisfying solution if we do not want to inhibit the development of this
technology and thus make Europe totally unattractive for AI developers107. Not only the provision
carries a lot of uncertainties (when exactly is a content online “lawfully available” to use? How
exactly to exercise the opt-out, or said differently, what is “an appropriate reservation by machine
readable means”? And who should be able to decide about this, the author or its derivative
rightholder?), but there is a risk that the opt-out is used as a new bargaining power to license the
use of existing works for training purposes. Given that generative AI systems are trained on an
immense number of works to be efficiently working, this would quickly turn out to be a licensing
nightmare for every AI developer in order to get all the relevant authorizations108. Also, one should
be wary of the fact that the author will not necessarily benefit directly from this situation as it will

107
See in this sense also M. Senftleben, “Generative AI and Author Remuneration”, supra note 81.
108
See in this sense J. Vesala, “Developing Artificial Intelligence-Based Content Creation: Are EU Copyright and
Antitrust Law Fit for Purpose?”, IIC 2023, Vol. 54, Issue 5, p. 351; C.J. Craig, “The AI-Copyright challenge: tech
neutrality, authorship and the public interest”, supra note 68 at p. 152, underlining that “given the sheer volume of text
and data mined to effectively train a sophisticated AI, limiting or foreclosing the use of copyright-protected works in
such processes in the absence of permission from the right holder places an enormous burden on AI-research and
development. Moreover, it produces de facto barriers to certain kind of AI projects, differentially disadvantages
anything but the most well-resourced AI researchers and exacerbates the built-in biases and discriminatory effects of
AI systems”.

28
likely be the big rightholders that will license the uses, with the authors having to (re)negotiate
successfully with their producers to get additional remunerations, which is not always an easy
task109.

Therefore, it might be interesting to reflect on a possible statutory remuneration to the benefit of

the author for the use of his works in the context of TDM activities for generative-AI purposes110.
In previous writings, we had proposed to replace the opt-out of article 4 CDSM by a statutory

109
For some interesting additional reflections on why authors would lose-out in such a scenario, see K. Trendacosta
and C. Doctorow, “AI Art Generators and the Online Image Market”, 3 April 2023, available at
https://www.eff.org/deeplinks/2023/04/ai-art-generators-and-online-image-market. Proposing however a consensus-
based remuneration system in the Asian context, see A. Dermawan and P. Mezei, “Artificial Intelligence and
Consensus-Based Remuneration Regime in Southeast Asia”, 7 Nov. 2023, available at ssrn:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4625850
110
See already C. Geiger “When the Robots (try to) Take Over: Of Artificial Intelligence, Authors, Creativity and
Copyright Protection”, in: F. Thouvenin, A. Peukert, T. Jäger and C. Geiger (eds.), “Innovation- Creation- Markets,
Festschrift Reto M. Hilty”, Springer, 2024 (forthcoming); C. Geiger and V. Iaia, “The Forgotten Creator: Towards a
Statutory Remuneration Right for Machine Learning of Generative AI”, supra note 27; in the context of the use of
journalistic content see C. Geiger and B.J. Jütte, “Designing Digital Constitutionalism: Copyright Exceptions and
Limitations as a Regulatory Framework for Media Freedom and the Right to Information Online”, in: M. Senftleben
et al. (eds.), “Cambridge Handbook of Media Law and Policy in Europe”, Cambridge University Press, 2024
(forthcoming); G. Frosio, “Should We Ban Generative AI, Incentivise It or Make it a Medium for Inclusive Creativity”,
in: E. Bonadio and C. Sganga (eds.), “A Research Agenda for EU Copyright Law”, Edward Elgar, 2024 (forthcoming).
For a similar proposal in its spirit, see M. Senftleben, “Generative AI and Author Remuneration”, supra note 81 at
p. 14. In his model, the statutory remuneration would however not be on the TDM use of protected works for AI
machine learning purposes, but it is “the literary and artistic output of generative AI systems” that serves “as a reference
point for a legal obligation to pay remuneration”. According to the author, focusing on an “output-oriented AI levy
system can be applied in a uniform manner to all providers of generative AI systems in the EU. In contrast to a
remuneration obligation focusing on the input dimension and AI training activities, the output-oriented levy approach
avoids the risk of disadvantages for EU high-tech industries. All providers of generative AI systems are equally
exposed to the levy payment obligation the moment they offer their products and services in the EU”. To counter
legal/doctrinal concerns and to give theoretical support to the proposal, Senftleben makes useful reference to the theory
of the domain public payant. If we agree with the outcome of this innovative proposal, in our view, an input-based
remuneration system caries however significant advantages. First, it gives legal certainty to AI developers as the
training of AI with protected works lies still in the grey zone with regard of its legality, in the EU but also outside: the
numerous lawsuits in the US challenging the legality of the use of copyrighted work to train AI demonstrates this.
Ultimately, it is not certain that the courts in the US (where the majority of cases are pending) will consider the machine
learning activities as a “fair use”. In any case, this will surely be a long judicial battle, carrying some uncertainties for
the small economic players, before a consistent case law is established under the fair use doctrine. Should the courts
declare the use unfair on the contrary, this would put the EU developers in a competitive advantage as a remuneration
right is always preferable then having to clear all the authorizations. The second argument in favor of a remunerated
exception is certainly that it is very much compatible with the European tradition of remunerated exceptions and that
there is an established practice and case law with regard to the distribution rules in favor of creators of these kind of
remunerations via collective management organisations. Finally, the idea of the domaine public payant, if admittedly
appealing in theory, might not benefit from a broad support; at policy level it could be more difficult to draw consensus
on a proposal based on a paid public domain. Advocates of a robust public domain might be favorable towards
ameliorating the remuneration situation of creators, less towards the idea of a domaine public payant.

29
remuneration for commercial TDM activities111, in order not to penalize start-ups developing useful
AI systems in the EU112. It needs to be recalled that AI systems cannot function without text and
data mining and therefore if we want to incentivize AI activities in the EU, TDM exceptions are
crucial. This certainly does not mean that all the uses should be free in all circumstances and in the
EU, there is a vast experience with the “permitted-but-paid”113 model of remunerated limitations
and exceptions (or, to use another terminology, “limitation-based remuneration rights”)114.

Thus, in analogy with the idea of a commercial TDM activity, a specific remuneration right to the
direct benefit of creators could be elaborated for the use of their work to train machines115, possibly
subjecting this right to mandatory collective management to make sure it can be rapidly
implemented (in the context of the exception for private copy for example, large sums are collected
and redistributed by collective management organizations to creators for the use of their works for
private purposes, via a relatively well functioning levy system116). In this context, it would also

111
Such as for example TDM brokers, proposing TDM searches on specific subjects to their customers against
remuneration. It is likely that these new services will increase in the future, and it could be considered that these
commercial TDM activities might be subject to payment of a statutory remuneration.
112
See the proposal in this sense C. Geiger, G. Frosio and O. Bulayenko, “Text and Data Mining in the Proposed
Copyright Reform: Making the EU Ready for an Age of Big Data?”, supra note 98 at p. 838.
113
For this terminology see J.C. Ginsburg, “Fair Use for Free, or Permitted-but-Paid?”, Berkeley Technology Law
Journal 2014, Vol. 29, Issue 3, p. 1446.
114
See C. Geiger, “Promoting Creativity through Copyright Limitations, Reflections on the Concept of Exclusivity in
Copyright Law”, Vanderbilt Journal of Entertainment & Technology Law 2010, Vol. 12, Issue 3, p. 515 sq.; C. Geiger
and O. Bulayenko, “Creating Statutory Remuneration Rights in Copyright law: What Policy Options under the
International Framework?”, in: H. Grosse Ruse-Khan and A. Metzger (eds.), “Intellectual Property Ordering Beyond
Borders”, Cambridge University Press 2022, p. 408 sq. (at 446).
115
For such a proposal in the context of machine learning see C. Geiger, “The Missing Goal-Scorers in the Artificial
Intelligence Team: Of Big Data, the Fundamental Right to Research and the failed Text and Data Mining Limitations
in the CSDM Directive”, supra note 99 at p. 392, in particular note 33. See also M. Kop, “The Right to Process Data
for Machine Learning Purposes in the EU”, supra note 99 at p. 7, proposing “the creation of an online one-stop-shop
clearinghouse with mandatory or statutory licensing for machine learning training datasets alike a pan-European, multi-
territorial collective rights agency”. More generally on the advantages of statutory licenses see R.M. Hilty,
“Verbotsrecht vs. Vergütungsanspruch: Suche nach Konsequenzen der tripolaren Interessenlage im Urheberrecht”, in
A. Ohly, M. Lehmann, T. Bodewig and T. Dreier (eds.), “Perspektiven des Geistigen Eigentums und
Wettbewerbsrechts. Festschrift für Gerhard Schricker zum 70. Geburtstag”, C.H. Beck, 2005, p. 325 sq.; C. Geiger,
“Promoting Creativity through Copyright Limitations, Reflections on the Concept of Exclusivity in Copyright Law”,
supra note 114 at p. 527, underlining i.a. that the earnings resulting from these rights can be in many cases much more
interesting for authors than the royalty payments they receive from contracting parties resulting from their exclusive
entitlements.
116
According to a study done by de Thuiskopie and WIPO, “International Survey on Private Copying”, 2016, pp. 15-
17, approx. 600 million euros per year was the average amount of private copying levies collected between 2007 and
2015 in 31 countries across the world covered by the survey. These global collections of private copying levies for all
rights holders since raised, according to a more recent study, totaling EUR 1,046 million in 2018 (CISAC, BIEM &
Stichting de Thuiskopie, “Private Copying Global Study 2020”, Nov. 2020, p. 8). They are a particular important
revenue source in Europe, where private copying remuneration systems amount for 1,019.4 million euros per year. For

30
make great sense to differentiate with regard to the purpose of the use and the works used in the
machine learning process117.

Of course, the amount of the remuneration to be paid needs to be monitored closely (and preferably
independently118) in order not to create a significant hurdle for start-ups and AI developers to
engage in this activity, which could be adjusted with regard to the economics of this sector and the
potential losses which could occur to the original creator (intuitively, but this would need to be
verified by empirical studies, training AI on highly successful works on the market will create
output that are likely to be commercially more successful, such as a song in “the style of” a
renowned artist). Also, and this should not be forgotten, authors already use (and might
increasingly use in the future) AI in their creative process as a tool for creativity, such as digital art
uses software or even more classically photography uses cameras and filters. Opposing
systematically AI systems and authors might not be a wise idea as they might very well cohabitate
in the future and support each other119.

the EU see also M. Kretschmer, “Private Copying and Fair Compensation: An empirical study of copyright levies in
Europe”, UK Intellectual Property Office, 2011, p. 7: “Following the Directive of 2001, total collection from levies on
copying media and equipment in the EU tripled, from about €170m to more than €500m per annum”. Admittedly,
these remunerations currently benefit both rightholders and authors, but nothing would prevent to increase the share
for creators and even to allocate the sums to them exclusively in the future in the context of a limitation-based
remuneration for machine learning purposes.
117
In this spirit, see J. Love, “We Need Smart Intellectual Property Laws for Artificial Intelligence”, Scientific
American, 7 August 2023, available at https://www.scientificamerican.com/article/we-need-smart-intellectual-
property-laws-for-artificial-intelligence/: “As laws and regulations emerge, care should be exercised to avoid a one-
size-fits-all approach, in which the rules that apply to recorded music or art also carry over to the scientific papers and
data used for medical research and development”. Research uses benefit from a strong fundamental rights justification
and might be treated differently (see C. Geiger and B.J. Jütte, “Conceptualizing a ‘Right to Research’ and its
Implications for Copyright Law, An International and European Perspective”, supra note 29); the statutory
remuneration solution for machine learning however might avoid many of the blocking issues that could emerge if
these uses were submitted to the exclusive right (in the EU, the TDM uses by research organisations are already
exempted by Art 3 TDM). What is evident however is that machine learning for research purposes should not be
subjected to exclusive control of publishers. As J. Love rightly points out, “it’s one thing to have the copyright holder
of a popular music recording opt out of a database; it’s another if an important scientific paper is left out over licensing
disputes”.
118
This could be for example done by a new EU independent copyright institution to be created, see in this sense C.
Geiger and N. Mangal, “Regulating Creativity Online: Proposal for an EU Copyright Institution”, GRUR Int. 2022,
Vol. 71, Issue 10, p. 933.
119
See in this sense a recent study of the University of Oxford entitled ‘AI and the Arts: How Machine Learning is
Changing Artistic Work’, where the scope of human/AI creative complementarity is examined through interview-
based case studies of the use of current AI techniques by creators of artistic works. The authors of the study conclude
that “despite the increased affordances of machine learning technologies, the relationship between artists and their
media remained essentially unchanged, as artists ultimately work to address human––rather than technical––
questions” (A. Ploin, R. Eynon, I. Hjorth and M.A. Osborne, “AI and the Arts: How Machine Learning is Changing
Artistic Work”, Report from the Creative Algorithmic Intelligence Research Project, Oxford Internet Institute,

31
Maybe the AI discussion could also be a good opportunity to reflect more generally on specific
remunerated exception for creative reuse that could englobe creative reuses of protected content,
as in the digital environment (in particular, but it is also valid in the analog world) creators have
increasingly incorporated protected elements in their creative process (sampling being a good
example)120. Of course, there might still be situations where the reuse of the work (by an AI or not)
leads to unwanted results. If an AI is trained with protected works to issue outputs that are
offensive, explicitly unwanted or carry inappropriate messages (racist, discriminatory, etc), there
should be room for authors to oppose that, but this could be dealt- as already discussed supra- with
as a moral right issue and not prejudice any practical but fair solution under the economic rights121.
The situation would not be radically different from a traditional moral rights violation, especially
if the original work is recognizable122.

VII. Conclusion and outlook

As we have seen, the development of AI generates a lot of questions123. How profoundly it will
affect our lives – and in particularly creativity and the cultural ecosystems- remains to be seen.
However, this is not the first -radical- technological (r)evolution that Mankind has faced. The legal
system- and in particularly the copyright system- (more or less) always managed to adapt. In any

University of Oxford, 2022, available at https://www.oii.ox.ac.uk/wp-content/uploads/2022/03/040222-AI-and-the-

Arts_FINAL.pdf).
120
See detailed on this proposal C. Geiger, “Freedom of Artistic Creativity and Copyright Law: A Compatible
Combination?”, UC Irvine Law Review 2018, Vol. 8, Issue 3, p. 413; C. Geiger, “Statutory Licenses as Enabler of
Creative Uses”, in: K.-C. Liu and R.M. Hilty (eds.), “Remuneration of Copyright Owners, Regulatory Challenges of
New Business Models”, Springer, 2017, p. 305 sq.
121
For moral rights issues raised by machine learning processes see J. Drexl, R.M. Hilty, et al., “Artificial Intelligence
and Intellectual Property Law”, supra note 76 at p. 12, stating in particular that “the right to integrity can pose
limitations to the training and to the creation of outputs by AI, namely, regarding the processing of protected works”.
122
Freedom of expression is not unlimited and moral rights can step in when a derivative work is offensive. See on
this issue, C. Geiger and E. Izyumenko, “Designing a Freedom of Expression-Compliant Framework for Moral Rights
in the EU: Challenges and Proposals”, supra note 32.
123
See, from a US copyright perspective, M.A. Lemley, “How Generative AI Turns Copyright Law on its Head”, 21
July 2023, available on SSRN: https://ssrn.com/abstract=4517702. See also in this sense the Memorandum Opinion of
Judge Howell in its decision of August 18, 2023, in Thaler v. Perlmuter et al., supra note 7: “Undoubtedly, we are
approaching new frontiers in copyright as artists put AI in their toolbox to be used in the generation of new visual and
other artistic works. The increased attenuation of human creativity from the actual generation of the final work will
prompt challenging questions regarding how much human input is necessary to qualify the user of an AI system as an
“author” of a generated work, the scope of the protection obtained over the resultant image, how to assess the originality
of AI-generated works where the systems may have been trained on unknown pre-existing works, how copyright might
best be used to incentivize creative works involving AI, and more”.

32
case, as in the past, the underlying human rights framework gives guidance and offers a workable
compass in navigating reforms of the current copyright system with regards to generative AI
systems. What is certain is that AI invites a much deeper reflection than the question of the
copyright issues of AI-generated works; and while lawyers are certainly concerned (as law frames
the society we live in), possible future solutions should mostly be taken according to economic,
philosophical, technological, artistic and ethical considerations. Certainly, we are only at the
beginning of an evolution. For now, as we have seen in this contribution, the copyrightability of
AI-generated outputs is to be considered with outmost care124 and, as it follows from a
fundamental/human rights perspective, only when AI is used as a technical tool for creators in their
creation process- meaning when they can serve a human author. However, we need also to be
careful that the development of AI systems is not inhibited as it can have a lot of beneficial aspects
if it is appropriately regulated125. Thus, we have proposed to replace the opt-out mechanism of
Article 4 CDSM by a TDM exception for creative purposes coupled with a statutory remuneration
to the benefit of authors only, in coherence with a proposal tabled in the past of a statutory
remuneration for creative uses126. As we have tried to demonstrate, this remunerated “right to train
the AI” can equally find support in the human rights framework. Of course, this proposal will need
to be developed, discussed, closely monitored and evaluated from a multidisciplinary perspective.
What is certain is that copyright law should secure a vibrant environment for culture and creativity
in the future. This can be done by regulating wisely these new technological environments, but this
also requires (finally) cherishing and putting the Human Author127 at the center of the copyright

124
See also H. Sun, “Redesigning Copyright Protection in the Era of Artificial Intelligence”, 107 Iowa L. Rev. 2022,
p. 1217, according to whom “AI works generated solely by autonomous AI systems should be placed in the public
domain without copyright protection”.
125
In particular O. Lobel, “The Law of AI for Good”, January 26 2023, San Diego Legal Studies Paper No. 23-001,
available on SSRN: https://ssrn.com/abstract=4338862. Regarding the benefits of Generative AI in the field of research
see S. Flynn, C. Geiger and J.P. Quintais, “Implementing User Rights for Research in the Field of Artificial
Intelligence: A Call for International Action”, EIPR 2020, Vol. 42, Issue. 7, p. 393; C. Geiger and B.J. Jütte, “The
Right to Research as Guarantor for Sustainability, Innovation and Justice in EU Copyright Law”, in: T. Pihlajarinne,
J. Mähönen and P. Upreti (eds.), “Rethinking the Role of Intellectual Property Rights in the Post Pandemic World: An
Integrated Framework of Sustainability, Innovation and Global Justice, Edward Elgar, 2023 (forthcoming), available
on SSRN: https://ssrn.com/abstract=4140627; Sean Flynn et al., “Legal reform to enhance global text and data mining
research”, Science 2022, Vol. 378, Issue 6623, p. 951.
126
See C. Geiger, “Statutory Licenses as Enabler of Creative Uses”, supra note 120. More generally on statutory
remunerations rights as a workable compromise solution in the digital environment, see C. Geiger and O. Bulayenko,
“Creating Statutory Remuneration Rights in Copyright law: What Policy Options under the International
Framework?”, supra note 114 at p. 408 sq.
127
See in this sense also D. Gervais, “The Human Cause”, supra note 35 at p. 38, “If we refuse to take the position that
the focus of IP law is human creativity and innovation, what will be left for us to do? Who will be the great creators
of tomorrow who will help us to understand ad shape our world if machines are the artists, novelists, and journalists?”

33
system (and not only the copyright industries). Doing this, we might be able to have in the future
AI-systems that serve creators and creativity, and not the other way around.

34