Artificial Intelligence and Copyright Issues
With the rapid development of artificial intelligence (AI) systems, the concepts of work and creativity have become one of the most debated areas of law. In Turkish law, a work is defined as "any kind of intellectual and artistic product bearing the distinctive characteristics of its owner" according to Article 1/Ba of the Law on Intellectual and Artistic Works (FSEK). To what extent do texts, images, music, or code fragments produced by AI fall under this definition? This question has necessitated a reinterpretation of the concept of "creative human contribution" from the perspective of copyright law
2. The Issue of Authorship in Artificial Intelligence Production
2.1. Fully Autonomous Production
When an artificial intelligence system creates a work without any human intervention, it is not considered a "work" within the meaning of the Turkish Copyright Law because it lacks human creativity . Therefore, copyright protection does not arise ; the produced content becomes public domain . The decisions of the US Copyright Office and the UK IPO are also in the same direction.
2.2. Production with Human Intervention
Products that reflect the "characteristics" of the person who directs the AI with instructions , designs the prompt , edits the output , or makes the selection may constitute a work of art. In this case, copyright belongs to the person who created the prompt or the person managing the project. Theories of co-authorship or creation of a work in favor of the employer can be applied under Articles 8 and 9 of the Turkish Copyright Law .
2.3. Prompt and Model Ownership in Artificial Intelligence Production
A prompt, that is, a set of instructions given to the AI, can be protected by copyright when it is a "creative command that directs the creation of the work."
However, technical prompts (e.g., "generate image style X") are generally not covered by copyright.
On the other hand, ownership rights over the AI model contractually .
-
The company that developed the model → the author,
-
The user obtains a license or derivative rights arising from this contract.
3. Copyright Issues During Data Set and Training Phases
Training AI models uses enormous amounts of data.
A significant portion of this data is protected by copyright.
3.1. Unauthorized Data Use
If the data consists of protected content such as news, images, or music, reproduction and processing may constitute a violation of Articles 22 and 21 of the Turkish Copyright Law.
Therefore, open-licensed (CC, MIT, etc.) datasets or collective licensing agreements .
3.2. Data Mining Exception
The "data mining exception" introduced by the European Union's 2019 Directive is not included in the Turkish Copyright Law.
Therefore, the use of copyrighted data for AI training in Turkey must be based on explicit consent or legitimate interest.
3.3. Personal Data Aspect
If the dataset contains personal data, processing conditions and anonymization obligations arise in accordance with Articles 5-6 of the Turkish Personal Data Protection Law (KVKK) . Otherwise, both copyright infringement and data breach may occur simultaneously.
4. Protection and Licensing of AI-Generated Productions
4.1. Maintainability of AI Output
The criterion of "bearing the characteristics of its owner" in Article 1/B of the Turkish Copyright Law is the most critical point for AI outputs.
If human control and creative contribution are present, it acquires the status of a work.
In this case, the rights holder is human; AI itself is not a rights holder.
4.2. Contractual Protection
The terms of use (usually EULA or Terms of Service) of AI platforms specify the licensing status of the outputs.
Some transfer full ownership of the output to the user, while others grant a co-license.
In Turkey, these agreements are valid under Articles 26-27 of the Turkish Code of Obligations based on the principle of "freedom of will"; however, fair terms and reasonable information obligations are required.
5. The Aspect of Proof and Dispute
Timestamps , blockchain records , log data , and metadata are crucial for determining the source of AI-generated content . In disputes, the answer to the question "who has the creative contribution?" is provided through a combination of technical evidence and contractual documentation.
6. Comparative Approaches to Law and Justice
-
USA: Emphasis on "human authorship requirement"; fully autonomous AI output is not protected.
-
EU: The author is solely human; contractual rights are possible for AI output.
-
United Kingdom: Copyright Designs and Patents Act, p. 9(3) → “The rights to works created by a computer belong to the person who created the work.”
This approach is not a precedent in Turkish law, but it can shed light on judicial precedents.
7. Fairness and Policy Assessment
Leaving artificial intelligence's creations completely unprotected, even without human input, can hinder technological innovation.
However, granting excessive rights also restricts access to information.
Therefore, the solution contractual rights regulations, collective licensing systems , and transparent data policies .
8. Conclusion
The relationship between artificial intelligence and copyright is not only a legal issue, but also an ethical and economic one.
The necessary steps for Turkish law in the near future include:
-
Adding a data mining exception to the FSEK ( Law on Intellectual and Artistic Works ),
-
Preparation of guidelines on the contractual protection of AI products
-
This involves the establishment of technological ethical standards by the Ministry of Culture and Tourism and the Personal Data Protection Authority (KVKK) .
Ultimately, AI is a new creator with whom the legal system must coexist in harmony; the task of the law is to draw the boundaries of this creativity so that it is compatible with human dignity and rights.
About Post Author
