WHO OWNS THE COPYRIGHT FOR WORKS GENERATED BY ARTIFICIAL INTELLIGENCE? CURRENT DEVELOPMENTS AND LEGAL DISCUSSIONS

Entrance

The Digital Transformation of Creativity and the Test of Law

Throughout human history, "creativity" has been considered a uniquely human, biological and mental faculty. The development of art and science has been based on the idea of ​​humans reflecting their emotions, thoughts, and aesthetic preferences to the outside world through a "work of art." Intellectual property law, the legal extension of this philosophical foundation, has centered on the relationship between the work and its creator, the "human," endowing the author with extensive material and moral rights. However, today, the autonomous creation of paintings, compositions, and the production of technical texts by generative AI systems has transformed the question of "who is the author?" from a philosophical debate into a central legal issue in the global economy.

By 2026, the interplay between artificial intelligence and legal systems will have evolved from a process of passively following technology to one of redefining legal rules themselves. The position that traditional regimes protecting intellectual property will take in the face of outputs that do not involve "human will" concerns not only the creators but also the companies that have made massive investments in these technologies.

DEVELOPMENT

1. The Concept of "Work" and the Criterion of "Owner's Uniqueness" from a Legal Perspective

The Turkish Law No. 5846 on Intellectual and Artistic Works (FSEK), the fundamental legislation regulating copyright in the Turkish legal system, requires two essential conditions that must be met cumulatively for a work to be legally protected. Firstly, the work must fall into one of the limited categories of works listed in the law (science and literature, music, fine arts, cinema); secondly, it must bear the subjective element of "the distinctiveness of its owner." In legal doctrine, "distinctiveness" refers to the author's personal stamp, their original creative genius, and intellectual effort. In other words, the resulting work must be a reflection of its creator's intellectual personality. This is precisely where the debate on artificial intelligence copyright hinges: How will data outputs that are not filtered through the human mind but are merely a product of algorithmic probabilities meet the "distinctiveness" criterion?

an algorithm processes billions of data points and produces an image or text based on statistical probabilities, it is quite difficult to find a legally defined "human quality" in that output. Traditional legal doctrine tends to position artificial intelligence as merely a tool (just like a brush, typewriter, or camera). According to this "tool theory," if artificial intelligence is passive technical hardware reflecting the user's will, then the copyright of the resulting product belongs to the user. However, as the autonomy and machine learning capabilities of artificial intelligence increase, this "tool" quality is seriously undermined. If an artificial intelligence system creates its own original composition without direct user intervention, the law struggles to speak of the existence of a "work" at this point. This is because, in the current Turkish legal system, the author of a work is the natural person who created it.

2. Investment Protection and the “Sui Generis” (Unique) Rights Model

By 2026, the most widely accepted approach in legal doctrine, and one that will guide draft legislation, "Sui Generis" rights category for artificial intelligence products. While traditional copyright protection aims to reward an individual's "creative genius," the sui generis rights model aims to protect the financial investment, technological infrastructure, and labor invested in producing that content. This model is particularly critical for protecting the rights of companies that train large language models (LLMs).

This proposed new model for AI outputs envisions protection similar to the "database creator's rights" already existing in our legal system. The aim here is not to bestow the status of a "work of art" upon the machine output, but to protect the enormous processing power and data set investment used to produce that product. This model plans to grant the right to file lawsuits against third parties who copy the AI ​​output, based on "unfair competition" or "investment protection." However, the duration of this protection should not be as long as traditional copyright protection (70 years), a more reasonable period of 5 to 15 years to avoid hindering innovation.

3. Ownership Scenarios in Artificial Intelligence Outputs

There are three main theories worldwide regarding the ownership of a product created by artificial intelligence. Each theory has both economic and ethical defenses:

I. User's (Prompt Owner's) Ownership Rights

In this scenario, the person who directs the artificial intelligence, gives the prompts, and defines the parameters is considered the creator of the work. Proponents cite the debates surrounding the invention of the camera as an example. Just as the person who presses the shutter owns the photograph, the person who shapes the outcome by giving the correct commands to the artificial intelligence should also be considered the creator of the work. However, the threshold here is "creative control." The legal status of a user who produces an output with just a single word should not be the same as that of a user who writes commands containing pages of technical details.

II. Software Developer's Ownership Rights

It is argued that the individuals or companies that code the algorithm and train the artificial intelligence should have ownership rights to all products resulting from this system. According to this view, the creative potential of artificial intelligence depends entirely on the genius of the programmer. However, this approach is criticized on the grounds that it can lead to disproportionate results, such as the person who wrote the programming language having a share in every book written in that program.

III. Becoming Public Property

It is argued that if artificial intelligence is the primary element in the creation of a work, and human intervention is minimal, then that product should become directly in the public domain without being subject to any intellectual property protection. This scenario is viewed with concern by creative industries because it could lead to a "free-riding" problem in the digital economy.

4. Artificial Intelligence Training and Copyright Infringements

The other side of the coin in the AI ​​copyright debate is the massive datasets that enable these systems to produce autonomous outputs. Generative AI models "data scraping" . In this process, digital libraries, art portals, and news sites on the internet become data sources; however, the artists and writers who are the original owners of these resources do not participate in the process with their consent.

Discussions on Violation of the Right to Adaptation and Reproduction

Artists and rights holders argue that using their works to train an algorithm constitutes a violation of the "right of reproduction" (Article 22) and "right of adaptation" (Article 21) under Law No. 5846 on Intellectual and Artistic Works .

• Regarding Reproduction Rights: Creating digital copies of works during AI training constitutes technical reproduction. While technology giants claim this is for temporary storage or analysis purposes, the property-focused view emphasizes that this copying is subject to the consent of the copyright holder.

• Regarding Adaptation Rights: The production of new outputs by artificial intelligence through "learning" from existing works is described by some legal experts as "adaptation at the stylistic level" of the work.

The 2026 EU Artificial Intelligence Act (AI Act) and the Transparency Revolution

The European Union's Artificial Intelligence Act (AI Act) , which will fully come into effect in 2026 , has made the principle of "transparency" a mandatory standard to address this legal complexity. AI developers are now required to publicly disclose a detailed list of which copyrighted data they used to train their models. The "Opt-Out" right granted to copyright holders allows artists to request the removal of their work from training kits.

5. The Fate of Moral Rights: Can a Machine Feel “Dignity”?

Moral rights regulated under the Turkish Copyright Law (FSEK) encompass rights such as making the work available to the public, having one's name indicated, and preventing alterations to the work. These rights protect the personal connection between the author and the work. In the case of artificial intelligence outputs, establishing this connection is impossible. It is legally technically impossible for an algorithm to claim that "my honor has been damaged due to alterations to my work." Therefore, it is highly likely that future legal regulations will only provide a limited protection regime for artificial intelligence outputs, focusing solely on "financial rights" (commercial use rights).

6. Future Projections for Turkish Law

In Türkiye, the Copyright Law needs revision to adapt to technological advancements. Unless Article 1 is amended or the Supreme Court makes a radical change in its jurisprudence, registering "pure" artificial intelligence products will be difficult. However, in complex works where the user utilizes artificial intelligence as an instrument and makes manual adjustments to the outputs, the user's authorship may be recognized.

Conclusion

The debate over ownership rights to works produced by artificial intelligence represents a process in which the definition of creativity is shifting from the human mind to the machine processor. By 2026, the legal system will have proven to be far from a static structure, forcing AI to evolve from a "human-centered" approach to one focused on "creative control and investment." In the future, the value of content will be determined not only by who produced it, but also by the technological investment and intellectual effort expended to create it. Intellectual property law will adapt to this digital revolution, continuing to protect both the speed of machines and the genius of humans under the same legal framework.

FREQUENTLY ASKED QUESTIONS

1. Can I add a "Copyright Belongs to Me" note to an image I generated using AI? Under the current legal system, this is risky. If there is no significant human intervention (editing, prompt elaboration) on the image, courts may determine that the content is not a copyrighted work. However, your commercial usage rights may be reserved under the platform agreement.

2. Would an article I wrote using ChatGPT be considered "plagiarism" from an academic standpoint? Yes, academically and ethically, it is a violation unless the use of artificial intelligence is explicitly stated. From a legal perspective, it is difficult to consider it a copyright infringement unless it involves copying the original statements of another author; however, protecting the authorship of the work is also difficult under current laws.

3. Is it a crime if artificial intelligence software imitates an artist's style? In law, "style" or "manner" is not protected. However, if it directly imitates the original composition or brushstroke arrangement in a particular work by an artist, this may constitute a violation under the Law on Intellectual and Artistic Works.

4. In AI-related copyright cases, who bears the burden of proof? The burden of proof as to whether content was created "by a human" or "by artificial intelligence" rests with the person claiming rights to that content. Digital watermarks and creation logs are the most important evidence in this regard.

5. Can I use ChatGPT or Midjourney outputs for commercial purposes? This depends on the respective platform's Terms of Service. Most platforms transfer usage rights to the user with paid subscriptions; however, this may not legally guarantee full "authorship" of the work.