AI inventorship open for debate in Turkey amid global DABUS rulings

“In recent legal debates, the patentability of AI-generated inventions has been contentious. Historically, only humans were considered inventors, anchored in notions of personhood and intellect. Yet, in a bold move, South Africa and Australia's Federal Court recognised AI, specifically DABUS, as a potential inventor in 2021, defying conventional views. Proponents argue that this accommodates modern innovation trends, while sceptics raise concerns about ownership rights, potential inundation of trivial patents and diminished human oversight. With patent laws varying globally, the landscape remains dynamic. As AI's role in R&D grows, more countries could re-evaluate their stances, reshaping the IP sphere.”

That introductory paragraph was created by an AI.

Today, AI is at the forefront of innovation as it seamlessly blends into roles traditionally held by humans. At some stage this is highly likely to include inventions too.

The concept of accepting AI as an inventor is relatively novel, with different countries adopting varying stances. Based on WIPO data, Figure 1 shows the different AI policies that leading economies are adopting, which primarily encompass database rights and national strategies.

The discussion about whether an AI system can be deemed an inventor began after the development of DABUS, an AI system created by Dr Stephen Thaler.

DABUS has claimed to have invented two novel products:

• a food container that uses fractal geometry, which enables rapid reheating; and
• a flashing beacon for attracting emergency attention.

Thaler and Professor Ryan Abbott and his team of attorneys filied applications at patent offices worldwide, aiming to establish precedents with regard to whether AI can be an inventor. This project, known as the Artificial Inventor Project, is intended to promote dialogue about the social, economic and legal impact of frontier technologies and generate stakeholder guidance on the protectability of AI-generated output.

Turkey’s position on AI

In Turkish patent law, the ‘actual ownership of the right’ principle dictates that the individual who creates or develops intellectual property should be recognised as the rightful owner. Article 109/1 of the Industrial Property Law states that "the right to obtain a patent belongs to the inventor or his legal successors, and it is possible to assign this right to others". Evidently, Turkish patent law does not permit AI to be designated as an inventor on patent applications.

For an invention to be patented, it must:

• be novel;
• involve an inventive step; and
• be applicable to industry.

These will also be considered in the patentability assessment of an invention developed by AI. The legal nature of the inventor is not listed among the patentability criteria, as legislators did not consider the possibility of AI being an inventor when drafting these provisions.

While the current stance is maintained, it is vital to bear in mind that no legal precedents address this matter – as seen in Thaler's cases – and the Turkish Patent and Trademark Office (TPTO) has not received any applications designating AI as the inventor. Consequently, this matter remains open for debate and contemplation.

Nevertheless, it is expected that Turkey’s approach would be similar to that of the EPO. Indeed, Turkey is a signatory of the EPC, and was also designated in the DABUS filings. Therefore, if Thaler’s patent application was registered before the EPO, it would also be registered in Turkey.

For Turkey – and many other countries – the current challenge revolves around the potential outcomes if AI were allowed to be recognised as an inventor. Therefore, the first step is to ascertain the legal status of AI and resolve this issue, potentially by amending the Industrial Property Law or establishing legal precedents.

There is no regulation in Turkey for AI yet. The Turkish Civil Code regulates natural and legal persons. According to the code, every individual has the capacity to exercise rights and all human beings are equal in terms of their rights and obligations within the legal system. It also regulates the capacity (and conditions thereof) to act, and accordingly, a person who has the capacity to act can acquire rights and incur debts through their actions. Legal entities, on the other hand, are subject to all rights and obligations other than those inherently related to human beings (eg, sex, age and kinship), and legal entities acquire the capacity to act by having the necessary organs.

Some Turkish scholars argue that AI should be considered a legal person. However, this does not mean a real person as in a human being, but rather a legal entity (eg, an association) since it is similar to the relationship between an association and its board members.

However, according to another perspective, Roman law may be an option for solving AI's legal personality problem. This idea suggests that the status of enslaved people under Roman law can be considered and applied as an example of the status of AI in terms of holding economic and IP rights. Under Roman law and the peculium system, enslaved people were able to have limited – yet legal – relations with third parties, even if they were considered another’s ‘property’.

Similarly, AI would be able to hold IP rights, and only for that limited value could it be held responsible for the legal consequences of its actions. It could be named as an inventor if the other requirements for patentability were fulfilled in the invention. In this way, the issue of who would own the rights and the responsibility matter could be solved, at least for now.