Approaches to IP protection for works generated by artificial intelligence

European standards

Works that are completely generated by artificial intelligence are not protected by copyright in the European Union. This is based on the principle that since there is no human intervention in the process of creation, the work cannot be considered original.

In the case of Infopaq International A/S v. Danske Dagblades Forening (C-5/08, 2009), the Court of Justice of the European Union stated that a work is considered original if it is “the author’s own intellectual work” and reflects their “personal contribution.” This means that human involvement in the creative process is essential. For works created solely by artificial intelligence, the absence of human intervention likely prevents them from meeting the originality criterion established by EU standards.

 The validity of this approach is supported by the conclusions drawn in the case of Levola Hengelo B.V. v. Smilde Foods B.V. (C-310/17, 2018). This case underscores the importance of a creator’s deliberate choices in recognizing a work as their own. When a person utilizes AI as a creative tool—such as by setting parameters, selecting results, or making significant adjustments—the resulting work may be eligible for copyright protection. This concept aligns with the idea of an “auxiliary tool,” where the creator maintains control over the creative process.

This perspective is further reinforced by EU scientific research, notably in publications from the European Commission, as well as by national court practices. For example, in 2024, the Municipal Court of Prague ruled that a complex creative request could grant a person copyright for a work generated using DALL-E. Consequently, the current legal status of works that are entirely generated by AI remains unregulated.

The classification of objects created by artificial intelligence as works was addressed in the European Commission final report on trends and developments in AI published in 2020. The report outlines that an AI-generated object can be considered a work if it meets the following criteria:

1. It pertains to the literary, scientific, or artistic sphere;
2. It results from human intellectual activity;
3. It reflects the creative choices of the individual who expressed them in the object.

The countries of the European Union are actively working on a set of legal frameworks that will address copyright issues and establish the boundaries of legal protection for works created using artificial intelligence (AI). This effort is part of the proposed AI Act and AI Code. Currently, discussions are ongoing, and the process of finalizing the act and the accompanying code is still in progress.

 According to the latest drafts of the AI Act, the status of works generated by AI remains unchanged. However, the drafts emphasize the importance of ensuring transparency and accountability in AI systems that produce content. Additionally, they outline mechanisms for licensing and regulating the use of these works, which will help clarify the relationships between AI technology developers and content users.

Experience from across the ocean

In the United States, the situation is similar: works may be eligible for copyright protection only if they are created by a human being—that is, when there is a human contribution and a creative element involved. In February 2023, the U.S. Copyright Office issued a decision stating that the images in Kris Kashtanova’s graphic novel “Zarya of the Dawn,” which were generated using the Midjourney AI system, are not eligible for copyright protection. The author may claim protection only for those parts of the book that she wrote and arranged but not for the AI-generated images. U.S. case law also confirms that only natural persons can be considered authors.

Following this case, in March 2023, the Copyright Office released Guidance: Works Containing Material Generated by Artificial Intelligence. The document states that individuals using AI technologies to generate a work may only claim copyright protection for their own contribution and must identify in the standard application the part of the work they claim authorship over.

In August 2023, U.S. District Judge Beryl Howell ruled in Stephen Thaler’s case that works created solely by AI without human involvement are not eligible for copyright protection. The judge noted that “human authorship is a bedrock requirement of copyright.”

In addition, a series of class action lawsuits filed by Sarah Andersen, Grzegorz Rutkowski, Jingna Zhang, Kelly McKernan, and others against GitHub Copilot (an AI coding assistant built on open-source software), Stability AI, DeviantArt, Midjourney, Runway AI, and Google (Imagen tool), have spurred the U.S. Copyright Office to draft and publish a policy in July 2024. This draft aims to regulate copyright issues related to AI-generated works by creating a unified approach to identifying such works, determining registration timelines, and establishing mechanisms for resolving copyright violations involving AI.

Protection in Ukraine

In Ukraine, efforts have also been made to regulate the legal status of works entirely generated by AI. The new Law “On Copyright and Related Rights” introduces a sui generis protection regime for non-original works generated by computer programs.

Article 33 of the Law of Ukraine “On Copyright and Related Rights, outlines the conditions under which such works are granted protection: the work must be distinct from pre-existing works, generated by a computer program as a result of its technical operation, and without direct human involvement. This means that the individual does not use any design systems or graphic software to create the object, and their role is limited to merely activating the program. There is no creative element, and the individual does not make intellectual decisions in the creation process. The psychological or emotional state of the individual has no impact on the outcome. As a result, such works are not considered copyrightable. Introducing a sui generis right grants these works legal protection where traditional IP protections would not apply.

Another distinguishing feature of such non-original AI-generated works is the duration of protection. According to Article 33 of the Law of Ukraine “On Copyright and Related Rights, protection lasts 25 years from January 1 of the year following the year in which the work was generated. In contrast, the duration of copyright protection for works created by natural persons is longer: 70 years from January 1 of the year following the author’s death or, in cases of anonymous or pseudonymous works, from the year of publication or the death of the last surviving co-author (Articles 31(2), 31(3), and 31(4)).

Conclusion

Thus, the question of legal protection for fully AI-generated works remains unresolved. Both in the EU and the U.S., the absence of human creativity precludes classical copyright protection. Ukraine’s approach stands out by introducing a sui generis legal regime for non-original works, providing a minimum level of protection. Nevertheless, this represents a compromise rather than a comprehensive solution.

It may be assumed that states will be forced to seek a balance between protecting human rights, advancing AI technologies, and preventing misuse.

Published by UNBA

Link to the article