Who owns the copyright to software generated using artificial intelligence?
1. Introduction
Today, artificial intelligence (AI) technologies play an increasingly important role in software development processes. In many software projects, all or a significant portion of the code is now generated by AI algorithms without direct intervention from human programmers. This situation has made the question of "Who owns the copyright to software generated by AI?" a subject of debate in the legal world.
The Turkish Law on Intellectual and Artistic Works (FSEK) and international intellectual property law norms recognize the author as "the person who has expended intellectual effort on the work." However, software generated by artificial intelligence is creating a new field that challenges this definition.
2. Artificial Intelligence and the Current Framework in Software Copyright Law
2.1. Definition of Work and Author According to the Copyright Law
According to FSEK m.1/B, a work is "any kind of intellectual and artistic product that bears the characteristics of its owner and is considered as a work of science, literature, music, fine arts or cinema". Software works fall into the category of works of science and literature within the scope of m.2/1 (1).
Here are the important criteria:
-
Intellectual effort: This must be the mental labor of the person who created the work.
-
Originality: The work must bear the original contribution of its author.
2.2. Codes Generated by Artificial Intelligence
Artificial intelligence generates code by learning from existing data and using algorithms. Since artificial intelligence has no will or personality in the production process, it cannot be legally considered the "author." Therefore, copyright is determined based on one of the following possibilities:
-
Code generated entirely through human intervention → The author could be the person using artificial intelligence.
-
Automatically generated code (with minimal human input) → its status as a "work" under the Turkish Copyright Law becomes debatable.
-
Using code derived from existing works → Risks infringing on the rights of the source code owner.
3. Artificial Intelligence and Copyright in International Law
-
USA: The US Copyright Office announced in 2023 that "works not contributed by humans" will not be eligible for copyright protection.
-
EU: The European Union Copyright Directive seeks human creative input in the production of artificial intelligence.
-
England: According to Article 9(3) of the Copyright, Designs and Patents Act, the person who runs the relevant programme is considered the author of works produced by a computer.
These differences are important for software developers and companies in defining global strategies.
4. Case Study Analysis
Event 1: KodX Software Company
AK, a software developer at KodX, created key modules for a large financial software application using an AI code generation tool. Much of the code was automatically generated by the AI. However, AK edited and optimized the code snippets produced by the system.
-
Legal assessment: Since there is human contribution (editing, adaptation) here, AK's copyright arises. However, due to the employment relationship, this right may be deemed transferred to the employer, KodX (FSEK art.18/2).
Event 2: Independent Developer MY.
MY generated game engine code every 5 minutes using fully automated artificial intelligence software. No manual intervention was made to any part of the code.
-
Legal assessment: Since it lacks human creativity, it may not be considered a work under the Turkish Copyright Law. However, in some countries, such as the UK, the person running the software may be the rights holder.
5. Copyright Infringement Risks in Artificial Intelligence Production
Because artificial intelligence generates code by learning from existing datasets, there is a possibility that the generated code could be copied "exactly" or "similarly" from another work.
-
Risk 1: Reproducing the licensed code found in the dataset exactly as it is.
-
Risk 2: Use in violation of license terms (e.g., using GPL-licensed code in a commercial product).
-
Risk 3: Unintentionally creating a patent infringement.
6. Criteria for Determining Entitlement
-
The level of human involvement (elements of creative control and selection).
-
Technical intervention during the software development process.
-
Contracts: Copyright transfer provisions between employer and employee, and between client and software developer.
-
Licensing method: Open source or proprietary?
7. Protection of Rights and Legal Process
7.1. Methods of Detection and Proof
-
Code versioning records
-
AI usage log records
-
Digital timestamps indicating the creation date of the artwork
7.2. Available Legal Remedies
-
Lawsuit for the cessation and prevention of infringement pursuant to Articles 66 et seq. of the Law on Intellectual and Artistic Works
-
Claims for material and moral damages
-
Contractual lawsuit in case of license breach
-
The use and distribution of the code have been halted as a precautionary measure
8. Preventive Legal Strategies
-
Copyright analysis should be performed for every piece of code generated by artificial intelligence
-
Adding clauses to employment contracts that clarify ownership rights to works generated by AI
-
Compliance audit of open source licenses
-
Notarized code registration or registration with TURKPATENT, if required
9. Conclusion
AI-generated software is opening up new areas of debate in copyright law. Under Turkish law, human contribution is a prerequisite for authorship. However, different approaches exist in international law. Therefore, software companies and developers should plan their contracts and development processes taking into account both national and international regulations.
Gamze Akbulut, Law Faculty Student
About Post Author
