Pehlivan & Güven Law FirmPublished:
The dizzying progress of generative artificial intelligence technologies over the last few years has reopened the debate on intellectual property, one of the most deeply rooted concepts in law. The spread of systems capable of independently producing a text, an image or a technical solution is stretching the limits of concepts historically regarded as uniquely human, such as "authorship", "originality" and "inventorship". The entire architecture of intellectual property law was built to encourage and protect creations of the human mind; yet today, the outputs produced by machines often cannot be distinguished from human works in quality and appearance. This clearly shows that existing legal frameworks must adapt in order to remain current and effective. This article examines the position of AI-generated output under copyright and patent law in the light of international developments and assesses the current picture under Turkish law.
Artificial intelligence is an umbrella concept covering a wide variety of systems with different structures and functions. While for a long time there was no commonly agreed definition, the definition adopted by the OECD in 2023, which is also consistent with the European Union Artificial Intelligence Act, has largely filled this gap. As systems become increasingly autonomous and able to make independent decisions, the question arises whether they should be granted some form of legal personality. The "electronic personhood" proposal put forward in this context envisages attributing certain rights and obligations to AI, but it brings with it complex ethical and legal questions concerning liability, accountability and the protection of human interests.
By contrast, the "functional personhood" approach proposes to proceed not from an entity's inherent qualities but from the functions it performs within the legal framework (such as producing original works, making autonomous decisions or entering into contracts). Although this approach offers the possibility of flexible integration consistent with the evolving capabilities of AI, in any event a careful balance must be struck so as not to weaken the rights and responsibilities of human actors.
At the heart of copyright law lies the understanding that a work is an original product reflecting its author's own intellectual effort. It is precisely at this point that the integration of AI into production processes challenges the traditional concept of authorship. In academic literature, a significant number of experts favour denying protection to AI-generated output, on the grounds that the true purpose of copyright is to encourage human creativity.
Among the proponents of this view, Daniel Gervais argues that AI does not need the same incentives as humans and that machines cannot make the creative choices necessary for originality. Craig and Kerr argue that authorship is in essence tied to human communication and that attributing authorship to AI is therefore incompatible with the nature of the concept. James Grimmelmann takes a more radical position, stating that there is, at present, no such thing as a genuinely "computer-authored work". By contrast, Ginsburg and Budiardjo argue that even the most advanced systems are tools acting under human direction, and that attention should therefore shift from the question "can AI be an author?" to assessing the claim of human authorship in the specific case. Annemarie Bridy proposes a different route, arguing that AI-generated output could be assimilated into the existing copyright system through the "work made for hire" doctrine of US law.
There is no uniform position in comparative law. In the United States, the "monkey selfie" decision (Naruto v. Slater) and in particular the 2023 decision in Thaler v. Perlmutter take human authorship as the basic condition for copyright protection. Under this approach, works produced by AI are attributed neither to the programmer, nor to the system, nor to the user who entered the prompt; in the end, these outputs are treated as being in the public domain without protection. Nevertheless, since the Copyright Office does not examine of its own motion whether applications are AI-generated, there are examples in practice where an AI-generated image has been registered by attributing it to a human; however, registrations made by deliberately concealing the role of AI are regarded as fraud.
In the European Union, although the body of legislation does not expressly regulate the copyright status of AI-generated output, the criterion of "the author's own intellectual creation" adopted by the Court of Justice in the Infopaq decision is interpreted as requiring substantial human involvement for protection. Whether this criterion is met for AI-generated output is left to the discretion of each member state. German and French law likewise tie originality to a person's personal touch and intellectual effort, excluding from protection outputs generated automatically without a genuine personal contribution. The United Kingdom has produced a distinctive solution in CDPA s. 9(3): for computer-generated works with no human author, it grants authorship to "the person by whom the arrangements necessary for the creation of the work are undertaken" and provides a fifty-year term of protection for these works. Meanwhile in China, the Shenzhen court's decision to grant copyright protection to an article produced with Tencent's Dreamwriter software, on the grounds that the structure of the content was reasonable, its logic clear and that it had a certain originality, reveals a different orientation. This picture shows that widely differing policy choices are in force around the world in response to the same technological reality.
Patent law is built on the assumption that inventorship is a uniquely human endeavour. However, the emergence of systems capable of producing new and useful solutions without human intervention has shaken this assumption. Under the "Artificial Inventor Project" led by Dr Stephen Thaler, applications naming the AI called DABUS as the inventor were rejected in most jurisdictions, notably the US, the United Kingdom and the European Patent Office, on the grounds that existing laws require a human inventor. This process has triggered a global debate on whether patent law should evolve to cover non-human inventors.
The consequences of recognising AI as an inventor are profound. On the one hand, the scope of patentable inventions could widen and innovation could accelerate; on the other, complex ownership issues arise, such as who the patent will belong to and how it would be shared among the developer, the user and the system itself. As current legislation is not prepared to deal with situations where an invention is produced entirely autonomously, new legal doctrines need to be developed in this area.
There is no specific regulation in Turkish law on the protection of AI-generated output. As regards patents, Article 90/5 of the Industrial Property Law (SMK) requires the inventor to be named in the application, and Art. 109/1 provides that the right to apply for a patent belongs to the inventor or their successors. Although the word "person" does not expressly appear in these provisions, the structure of the law and the fact that those entitled to protection are listed as "persons" in Art. 3 are interpreted as requiring the inventor to be a human. Since AI systems are not yet recognised as persons, it does not appear possible under current legislation to name an AI as the inventor or the applicant.
In copyright law, too, the Law on Intellectual and Artistic Works rests on the understanding that the author of a work is the natural person who created it; originality is assessed by whether the work bears the personal imprint of its author. In this framework, outputs produced entirely by AI are difficult to protect as works to the extent that they do not bear a human imprint. In conclusion, there is a clear legal gap in Turkish law as regards both copyright and patent protection for AI-generated output.
The intersection of artificial intelligence and intellectual property law holds both great potential for innovation and serious legal uncertainty. It seems inevitable that cornerstone concepts such as authorship, originality and inventorship will have to be reassessed to address the reality of AI output. The path to follow in this process is to develop flexible and balanced regulations that recognise the contributions of technology without compromising the principles of justice, fairness and the protection of human creativity. Given the disputes that uncertainty over ownership of rights may cause, carefully structuring licence agreements, terms of use and content production processes with this in mind is an important precaution that rights holders and users alike should take now.
