Gökçe Ergün, Çağla Yargıç and Yaren Türe of Kılınç Law & Consulting report on how Turkish law views AI-generated content in the context of authorship and copyright
Artificial intelligence (AI) tools that, in simple terms, bring algorithm-based machine learning to mind, have started generating unpredictable outputs, especially with the advancement of cutting-edge systems like generative AI contributing to technological advancement. AI can generate content across a wide spectrum, ranging from texts to images, music to videos.
However, the widespread production of content in this manner has raised issues related to intellectual property law. As a matter of fact, the convergency of the content obtained through AI into a novel nature raises questions about the concepts of the work of art and its authorship.
This article provides an evaluation within the scope of intellectual property law of content created with AI, discussing the concept of authorship and whether AI-generated content qualifies as a work. This article also examines the position of AI-generated content in relation to copyright and who may be liable in the event of copyright infringement.
Concept of authorship and evaluation of AI-generated content within the scope of copyright
With the rise of trends initiated by ChatGPT, the use of open-source generative AI applications by the public has made AI-generated content a part of everyday life. This situation brings together disputes regarding intellectual property rights and leads to many legal discussions ranging from copyright infringements caused by the data sets used by AI to the claims of third parties by AI users who claim copyright over the content they have produced.
‘According to the wording of Law No. 5846 on Intellectual and Artistic Works, in order for something to qualify as a work, it must bear the characteristics of its author.’
Gökçe Ergün, Çağla Yargıç and Yaren Türe, Kılınç Law & Consulting
The most recent and prominent dispute regarding the qualification of content generated by AI as a work is undoubtedly the Thaler v Perlmutter case. In 2018, Dr Stephen Thaler submitted an application to the United States Copyright Office (USCO) for a work titled ‘A Recent Entrance to Paradise’, which was generated through the AI system called the ‘Creativity Machine’. He argued that the rejection of this application for formal registration was unlawful. The objection was dismissed by the USCO on the grounds that the work was created without any creative contribution from a human actor and therefore could not bear the intellectual property characteristic of a human author. The claim was eventually brought before the district court, which upheld the USCO’s decision in August 2023. However, the judge noted the increasing lack of human creativity in the final work and emphasised the need to assess how much human intervention is required for an AI user to be considered the author.
Another important point about the case is that since Thaler applied for the work to be considered a ‘work-made-for-hire’, this aspect was also examined, and was ruled that the author must be a human being even in works created on order.
On the other hand, during the same period, Thaler submitted a patent application for an invention generated by AI to the United Kingdom Intellectual Property Office (UKIPO), asserting in this file that he owned any output produced by the AI system because he owned the ownership of the AI system to circumvent the requirements related to the application procedure. The court interpreted this matter as such a doctrine that has never been applied where a property generates intellectual property. This assessment indicates that intellectual property rights holders would not be considered as AI users in the Anglo-Saxon doctrine.
Under Turkish law, the concept of intellectual property includes rights and authority over literary and artistic works, as well as patents, utility models and designs. According to the definition provided in Article 1/B of Law No. 5846 on Intellectual and Artistic Works (LIAW), a work is defined as ‘all kinds of intellectual and artistic products that possess the characteristics of their creator and are classified as works of science and literature, music, fine arts, or cinema’. According to the wording of this law, in order for something to qualify as a work, it must bear the characteristics of its author. According to the doctrinal opinion, it is argued that an intellectual work with creative qualities and the reflections of the characteristics, creativity and style of the author’s idea can be specific to real persons and therefore, under the LIAW, works can only be created by real persons.
The 11th Civil Chamber of the Supreme Court’s decision dated 4 February 2015, numbered 2014/16277E., 2015/1285K., is similar to this view in doctrine, stating that;
‘The author of a work LIAW, in accordance with Articles 1 and 2/3, is the person who creates it. Since the plaintiff TSE is a legal entity established by Law No. 132 and has a legal personality, it cannot be said that the plaintiff, who does not have a creative activity, is the author of the work. The plaintiff TSE has the right to use the financial rights of the publications subject to the lawsuit under LIAW 10/last and/or Article 18, as a matter of law, but it cannot be said that it has moral rights ownership.’
The court ruled that the author must be a real person who can provide the creative activity.
Copyright concept in AI-generated works and liability in case of copyright infringement
Copyrights cover the legal rights of the products created by the intellectual labour of the person, and these rights are protected under LIAW and no registration is required in Turkish law for the copyright to arise. The rights to intellectual and artistic works arise with the production of the relevant work and since the protection of copyrights is considered within the scope of human rights, it is also protected by Article 27 of the Universal Declaration of Human Rights.
‘Turkish law does not provide for a copyright for autonomous AI products. While AI generates new content based on existing data, whether this content is completely original or not is a matter of debate in itself.’
Gökçe Ergün, Çağla Yargıç and Yaren Türe, Kılınç Law & Consulting
When the intersection of AI and copyright in Turkish law is examined, it should be noted that Turkish law does not provide for a copyright for autonomous AI products. While AI generates new content based on existing data, whether this content is completely original or not is a matter of debate in itself. Moreover, who will be held liable in a possible legal dispute that may arise is another matter of debate.
Indeed, due to the fact that the algorithms underlying AI rely on extensive datasets, it is also possible that the content they generate through processing these datasets may lead to the infringement of third-party copyright. In this respect, considering that AI lacks legal personality, it remains uncertain to whom claims arising from the infringement will be directed.
AI applications and their use for content production in everyday life have created various uncertainties in legal areas. These uncertainties encompass determining legal status, the concept of authorship, the applicability of copyright to AI-generated content, and what rights can be claimed in the context of current developments. These uncertainties are not limited to the ones mentioned.
Although there are differences in each country, under Turkish law it is not possible for a product generated by AI to be qualified as a work and to benefit from the rights and protections enjoyed by the products having the characteristics of a work within the scope of the LIAW in today’s legislation. On the other hand, the literature on the relationship between AI and intellectual property is becoming widespread on the global stage, both in case law and in the guidelines of institutions. Therefore, with the advancing technology, it will be inevitable that changes will occur in our legislation for the law to keep pace with the current technology.