I. INTRODUCTION
Generative
artificial intelligence is fundamentally transforming the traditional
mechanisms of intellectual creation, making it imperative for copyright law to
adapt to this new technological reality. AI models capable of generating text,
images, music, videos and computer code are permanently reshaping artistic,
scientific and professional activities. While these technologies offer
significant time savings as well as new opportunities for creation and
innovation, they also give rise to numerous legal issues, many of which remain
largely unresolved.
Under
French and European Union law, copyright protects original intellectual and
artistic works that reflect the personality of their author. Such protection
arises automatically upon the creation of the work, without the need for any
registration, and extends to intellectual and artistic works regardless of
their type, form of expression, value or purpose. Accordingly, scientific
articles, novels, musical works, photographs, computer programs and graphic
designs may benefit from copyright protection provided that they meet the
requirement of originality. However, the assessment of originality is made on a
case-by-case basis, taking into account the specific circumstances of each
work. In particular, products limited merely to the communication of data or
factual information may not qualify as protected works where they do not embody
a sufficient degree of creative contribution.
However,
generative artificial intelligence calls into question certain fundamental
principles of the copyright regime. These systems are often trained on vast
datasets that may include copyright-protected works. Through text and data
mining techniques, they analyze online content to identify specific structures
and statistical patterns and can subsequently generate new content based on
these data. Under European Union law, this process is governed by “Directive
(EU) 2019/790 of 17 April 2019 on Copyright and Related Rights in the Digital
Single Market”. The Directive adopts a novel framework that departs
from the traditional principles of copyright law. Accordingly, although the use
of a work is, as a general rule, subject to the prior authorization of the
rights holder, text and data mining activities may be carried out without
obtaining such authorization where lawful access to the content is available,
provided that the rights holder has not expressly reserved or opted out of such
use. Accordingly, research organizations and private entities developing
artificial intelligence systems may, subject to certain conditions, use
protected works for the purpose of training their models.
Nevertheless,
this framework gives rise to significant legal uncertainties. In practice,
rights holders are often unaware of whether their works have been used in the
training of an AI model, while users of AI systems may be unable to determine
whether the content generated reproduces pre-existing works. This lack of
transparency significantly complicates the assessment of copyright infringement
and contributes to an increasing number of disputes. Indeed, in recent lawsuits
brought against certain AI providers, rights holders have alleged that
copyright-protected works were used without authorization during the training
process or that the generated content was sufficiently similar to existing
works to create a likelihood of confusion and thereby constitute copyright
infringement.
They
also point out that AI systems may infringe authors’ moral rights by analyzing
and transforming their works. In addition, operators that derive economic
benefit from protected works without obtaining authorization may potentially be
held liable under the provisions of unfair competition law relating to
parasitism (“parasitisme”).
Beyond
the debates surrounding training data, the legal status of AI-generated content
also remains uncertain. Under the current legal framework, AI cannot be
recognized as the author of a work, since copyright is based on the
intellectual creation of a natural person. Accordingly, content generated
entirely by AI will, as a general rule, not benefit from the copyright
protection afforded to intellectual and artistic works. By contrast, protection
is not necessarily excluded where AI is used as a tool to support human
creative activity. As with software such as Photoshop or Word, AI may
contribute to the creation of an original work where the user’s creative
contribution is sufficiently determinative. Today, determining whether such
human contribution is sufficient constitutes one of the central issues in the
ongoing debate over copyright law in the context of AI.
In
addition, an important issue arises concerning ownership of rights in content
generated with the assistance of AI. Although AI itself cannot hold rights,
companies that develop or operate these technologies may assert certain rights
over outputs generated by their systems, arguing that they have contributed to
the creative process. In this context, it remains debatable whether such rights
should belong exclusively to the user, to the developer of the AI system, or
whether a regime of joint ownership should be adopted. Positive law has yet to
provide a definitive solution to this issue, clearly highlighting the
disconnect between the rapid development of technological innovation and the
comparatively slower evolution of statutory regulation.
II. EUROPEAN
UNION’S APPROACH TO GENERATIVE ARTIFICIAL INTELLIGENCE: THE EUROPEAN
PARLIAMENT’S RESOLUTION OF 10 MARCH 2026
Identifying
the inadequacies of the existing legal framework, the European Parliament calls
for enhanced protection of rights holders, greater transparency obligations for
AI providers, clarification of the conditions under which protected works may
be used for the purpose of training AI models, and greater legal certainty
regarding the status of AI-generated content. The resolution reflects the
European Union’s intention to strike a balance between fostering technological
innovation and safeguarding the fundamental rights of authors and other rights
holders.
The
development of generative artificial intelligence is fundamentally challenging
the traditional foundations of copyright law. These systems, which are capable
of generating content such as text, images, music, and video, are often trained
on vast datasets that include copyright-protected works. This technological
development raises a fundamental legal question: does copyright law, designed
to protect human creativity, provide an adequate and appropriate legal
framework for works resulting from the automated processing of protected
content?
To
address these issues, on 10 March 2026, the European Parliament adopted a
resolution aimed at adapting EU law to new technological developments while
preserving the fundamental principles of copyright law. The Parliament
emphasizes that copyright and related rights guaranteed under, in particular, Article
17 of the Charter of Fundamental Rights of the European Union, continue
to apply in full in the digital environment. In this regard, it stresses that
authors and other rights holders should retain control over the use and
commercial exploitation of their works.
The
Parliament also notes that many AI providers have trained their models using
protected works on a large scale without obtaining prior authorization. To
address this issue, the resolution proposes requiring AI providers to maintain
a detailed inventory identifying the works used in training their models. It
further emphasizes that rights holders should remain free to authorize or
refuse the use of their works and that they should retain the right to
negotiate the terms and conditions under which their works may be used.
However,
the resolution preserves the existing exceptions provided for the benefit of
scientific research, education, and non-commercial innovation activities, while
also reiterating that any use for commercial purposes remains subject to the
authorization of rights holders.
With
regard to AI-generated content, the Parliament reiterates that only original
intellectual creations reflecting the personality of a human author may benefit
from copyright protection. Accordingly, it states that content generated
entirely by AI without any creative human intervention should not be protected
under copyright law and may instead fall within the public domain.
At the
same time, the resolution acknowledges that creative processes are becoming
increasingly hybrid in nature and that AI is, in many cases, used merely as an
assistive tool. Accordingly, determining the level of human contribution
required for copyright protection to apply will be one of the key legal issues
facing European courts.
III. COPYRIGHT PROTECTION FOR AI-GENERATED WORKS IN THE UNITED STATES:
THE DECISION OF THE UNITED STATES COURT OF APPEALS FOR THE DISTRICT OF COLUMBIA
CIRCUIT OF 2 MARCH 2026
The approach of the United States to the protection of works generated
by artificial intelligence likewise confirms the human-centered nature of
copyright law. In its decision of 2 March 2026, the United States Court of
Appeals for the District of Columbia Circuit held that a work created entirely
by artificial intelligence, without human creative intervention, is not
eligible for copyright protection under the “Copyright Act”. The
decision became final after the U.S. Supreme Court declined to review the
appeal.
In this
present case, the owner of an artificial intelligence system sought to register
an image generated entirely autonomously by the system and identified the AI
itself as the author of the work. The U.S. Copyright Office rejected the
application on the grounds that U.S. copyright law necessarily requires the
existence of a human author. This interpretation was upheld first by the
federal district court and subsequently by the appellate court and was
ultimately affirmed when the Supreme Court declined to review the case.
This
approach is based on a literal interpretation of the “Copyright Act”.
Indeed, many provisions of U.S. copyright regulations are premised on the
assumption that the author of a work is a natural person. In particular, the
provisions governing the duration of copyright based on the author’s lifetime,
as well as those concerning the transfer of economic rights by inheritance, are
clear manifestations of this assumption. Accordingly, an artificial
intelligence system that has no legal personality and lacks legal capacity
cannot be recognized as an author under U.S. law. As a result, creations
generated entirely autonomously by an AI system are considered to fall within
the public domain and may be freely used by anyone.
Although
there are various differences between the American and European legal systems,
both systems adhere to a common fundamental principle: Accordingly, the central
legal debate is no longer whether an AI can be recognized as the author of a
work, as judicial authorities today clearly answer this question in the
negative. The key issue is to determine the extent of human intervention
required for a creation produced with the assistance of AI to still qualify as
an intellectual or artistic work. For such a qualification to be made, the
user’s creative contribution must be sufficiently substantial and identifiable
to justify copyright protection.
IV. PROTECTION OF AI-ASSISTED WORKS UNDER GERMAN LAW: THE MUNICH COURT’S
DECISION OF 13 FEBRUARY 2026
The
decision rendered by the Munich Court on 13 February 2026 constitutes a
significant milestone in shaping the legal framework applicable to works
created with the assistance of generative artificial intelligence. Without
calling into question the traditional principles of copyright law, the German
court clarified the circumstances under which an AI-assisted creation may
qualify for copyright protection.
The
dispute concerned a lawsuit brought by a user who had created three logos using
multiple “prompts” through a generative AI system. The plaintiff alleged that
the logos had been reproduced without authorization and brought a copyright
infringement action. However, the Munich Court dismissed the action on the
grounds that the logos did not qualify as works eligible for copyright
protection. According to the Court, the plaintiff had failed to demonstrate
that her/his personal contribution had exerted a predominant human creative
influence over the final form of the images. Accordingly, even where detailed
“prompts” are used, the mere use of generative AI is not, in itself, sufficient
to confer authorship on an individual.
Accordingly,
for a product to benefit from copyright protection, it must constitute an
original intellectual creation reflecting the personality of its author and
resulting from free and creative choices. The application of this principle in
the context of generative AI demonstrates that the decisive factor is not the
use of AI itself, but rather the existence of a sufficiently significant and
objectively identifiable human contribution to the final result. Accordingly,
AI is regarded merely as a technical tool at the service of the creative
individual.
However,
the Court adopted an “in concreto” approach, assessing each creation at
issue separately and taking into account the specific circumstances of the
case. The Court acknowledged that human contribution may arise through the
progressive formulation of prompts, the selection of results generated by the
AI, or modifications made to the image after it has been generated. By
contrast, the time spent preparing the prompts, the length or complexity of the
prompts, the use of a paid version of the software, or the user’s personal
effort and investment were not, in themselves, considered sufficient to
establish the existence of an original creation.
The
Court does not establish a special category of works for AI-generated
creations; rather, it reiterates that copyright protection depends on
demonstrating that the user’s creative choices have genuinely shaped the final
form of the work.
The
decision also demonstrates that documenting and tracing the creative process
are becoming increasingly important. Accordingly, the protection of a work
created with the assistance of AI will depend on the author’s ability to
demonstrate the creative process through which the work was produced. In this
context, it is particularly important to keep records of the prompts used, the
different versions generated, subsequent modifications, and the artistic
choices made throughout the process. This approach is also consistent with the
practice of the U.S. Copyright Office, which requires human contribution to be
demonstrated in a concrete and detailed manner.
V. LEGAL STATUS OF AI-GENERATED WORKS UNDER TURKISH LAW
Under
Turkish law, there is currently no precedent-setting decision by the Court of
Cassation or the Constitutional Court concerning the copyright protection of
works generated by artificial intelligence. Accordingly, the issue is assessed
primarily within the framework of the provisions of Law Nr. 5846 on
Intellectual and Artistic Works (FSEK) and the views put forward in legal doctrine.
The
legal attribution of content generated by generative AI systems, the parties
who may hold rights in such content, and the means by which such rights may be
asserted are currently the subject of extensive debate.
One of
the most significant aspects of these debates concerns the issue of liability.
Where content generated by AI infringes the copyright of third parties, it
remains unclear against which individual or entity the rights holder should
direct its claim. Under the current legal framework, attributing the
infringement directly to the AI system itself does not appear to be legally
possible.
Under
Turkish law, AI cannot, in itself, be recognized as the author of a work.
Pursuant to Article 1/B of the Law on Intellectual and Artistic Works, a
work is defined as “any kind of intellectual and artistic product that bears
the characteristics of its author and falls within the categories of works of
science and literature, music, fine arts, or cinematographic works,” while
an author is defined as “the person who creates the work”. Although the
statutory text does not expressly provide that the author must necessarily be a
natural person, the prevailing view in legal doctrine is that authorship may be
attributed only to natural persons, since authorship is the result of creative
intellectual activity. Accordingly, AI systems, which lack legal personality
and creative intent, cannot acquire the status of an author.
On the
other hand, legal entities likewise cannot acquire the status of author under
the framework of the Law on Intellectual and Artistic Works. Even where a work
is created under the instructions and supervision of a legal entity within the
scope of an employment relationship, an employment contract, or a similar legal
relationship, authorship belongs to the natural person who created the work. By
contrast, pursuant to Article 18 of the Law on Intellectual and Artistic
Works, unless otherwise agreed by the parties, the authority to exercise
the economic rights in works created by civil servants, employees and workers
in the performance of their duties belongs to the person or institution
employing them. Accordingly, a legal entity may not be the author of a work but
may have the authority to exercise the economic rights therein within the scope
of the authority granted by law. Thus, what is acquired in such circumstances
is not authorship, but economic rights in the copyright.
As
regards whether content generated entirely autonomously by AI can benefit from
copyright protection, the prevailing view in Turkish legal scholarship is
negative. The main rationale is that, under the Law on Intellectual and
Artistic Works, a product must qualify as a work, reflect the author’s
individual character, and result from human intellectual creation in order to
be eligible for protection. Since content created entirely by AI, such as a
text or image, does not involve human creative contribution, it is generally
accepted that such content cannot benefit from copyright protection. This
approach is also broadly consistent with the approach adopted in the decision
rendered in the United States in 2026 and in the Munich Court’s 2026 decision.
By
contrast, a different legal assessment applies where AI is used merely as a
tool. According to the prevailing view in legal scholarship, where the creative
process is directed by a human and the decisive intellectual contribution to
the creation of the work is attributable to that person, the resulting products
may qualify as works under the Law on Intellectual and Artistic Works and
benefit from copyright protection. In such cases, AI remains a technical tool
supporting the creative process, while the person making the intellectual
choices that shape the final form of the work holds the status of author.
Accordingly,
creative interventions such as determining aesthetic preferences and selecting,
modifying, or enhancing outputs generated by AI are considered among the
factors that may demonstrate the existence of human contribution, and the
resulting work may therefore be protected under the Law on Intellectual and
Artistic Works. By contrast, where AI operates entirely autonomously and the
human makes no creative contribution to the creative process, authorship cannot
be established in respect of the resulting product.
Therefore,
merely writing a “prompt” does not, in itself, confer authorship. The
prevailing view in legal scholarship is that the act of writing a “prompt”
alone cannot be regarded as a creative intellectual activity. For copyright
protection to apply, beyond formulating the “prompt,” the user must make
creative choices, modifications, interventions, and aesthetic decisions that
influence the final appearance of the work. This approach is also broadly
consistent with the Munich Court’s decision of 13 February 2026, which
identifies human creative contribution as the decisive criterion.
In
addition, the prevailing view in legal scholarship is that the use of protected
works without obtaining the authorization of the rights holders during the
training of AI models may constitute an infringement of the reproduction right
under the Law on Intellectual and Artistic Works and may violate the economic
rights of authors.
VI. CONCLUSION
In
conclusion, although generative AI has reopened debate on the fundamental
concepts of copyright law, recent developments in both the European Union and
the United States, as well as the prevailing doctrine in Turkish law,
demonstrate that human creative contribution remains at the heart of copyright
protection. At the same time, significant uncertainties persist in a number of
areas, including the training of AI models, the legal status of generated
content, and the applicable liability regime. In light of the rapid pace of
technological development, establishing a balanced and predictable legal
framework that both promotes innovation and effectively safeguards the
interests of rights holders will be one of the most important legal policy
challenges in the period ahead.
Elif
Keskin, Law Student Intern
References:
2. United States Court of Appeals
for the District of Columbia Circuit, 2 March 2026
4. https://www.economie.gouv.fr/apie/quel-droit-dauteur-lere-de-lintelligence-artificielle-generative
5. https://dergipark.org.tr/tr/download/article-file/3785580
