With the rise of AI capable of creating texts, music, or even works of art, copyright is more than ever put to the test. These new technologies, which push the limits of creation, question the very concepts of originality and protection of works.
The emergence of AI in the world of creation arouses legitimate concerns among human artists and authors. The risk that their works be imitated, reproduced, or outdated by IA creations jeopardized the value of their work and could even slow down human artistic creation. In parallel, the databases on which AI are based often contain protected works, raising ethical and legal questions about the right of authors to control the use of their works as raw material for non -human creations. This debate highlights the need for a clear framework which requires, for example, transparency and consent obligations concerning the use of such works in AI systems.
Copyright is a set of rights conferred on the creators of original works of the mind. It applies to literary, artistic, musical, audiovisual, photographic works, as well as software. Copyright gives the author of exclusive prerogatives on the use of his work, in particular the right to reproduce, distribute, represent publicly, and modify his work.
Works generated by AI: Who is the author?
One of the main questions that arises when we speak of artificial intelligence and copyright is to know who can be considered the author of a work generated by one by AI.
L'Article L111-1 of the Intellectual Property Code provides in particular that the author of a work of the Spirit enjoys this work, by the mere fact of its creation, of an exclusive intangible property right. This right includes intellectual and moral attributes as well as heritage attributes, which are determined by books I and III of this code.
Take, for example, an AI that creates a painting or composes music. Traditionally, the law recognizes an individual as the author of a work. But in the case of a work generated by an AI, where is paternity?
For the moment, the law does not recognize the possibility that an AI itself can claim copyright, because it takes human intervention and aware to claim this right. A work generated independently by an AI cannot therefore be protected by copyright if no human can be identified as author.
The absence of recognition of a clearly identifiable author complicates the application of copyright to the works generated by AI. Indeed, the law is based on proof that the work reflects the imprint of the personality of its creator. For the works produced by AI, it is a question of demonstrating that the prompts used testify to this personal imprint. If this evidence is not established, the work cannot benefit from the protection of copyright. The real issue lies in the massification of the content generated and the use of existing works to cause AI, which raises problems of counterfeiting.
IA training and intellectual property: a complex relationship
Training is the automatic learning process by which an algorithm or an AI model improves its performance on a given task by analyzing data.
In order for the AI to create works, they must be formed from large amounts of data, often from works protected by copyright. Models like those used by writing or generation of images need to ingest enormous volumes of information, often including literary works, music or paintings.
The authors of these original works are generally not consulted, and even less remunerated for this use of their work. This raises a crucial question: Is it legal to use protected works without the authorization of the beneficiaries to cause AI? In law, this maneuver constitutes counterfeiting under theArticle L335-2 of the Intellectual Property Code And legal proceedings have already been brought against companies that use AI models without respecting the rights of creators.
In the United States, collective appeals are underway to determine whether the unauthorized use of works for the training of AI is a copyright violation. For example, in 2023, a group of artists launched legal action against Stability Ai,, MidjourneyAnd Devantartby saying that their works had been used without authorization to train generative AI models.
Originality, creativity and AI
For a work to be protected by copyright, it must be original, that is to say that it must reflect the personality and the creativity of its author. But when a work is generated by an AI, how to assess this “originality”?
Can a work produced by an algorithm be considered original enough to be protected?
This debate is crucial for artists, writers and creators in general. Some lawyers argue that the works generated by AI cannot be protected because they do not reflect a ” human creativity“. Others believe that as long as the person using AI plays a significant role in the process, some protection should apply. This aspect makes the link with what was mentioned above.
By a decision of October 11, 2023, the Prague Municipal Tribunal spoke on the possibility of granting copyright protection to the content generated by an AI system. The court concluded that an image generated by an AI does not constitute a work of the mind and, therefore, cannot claim protection by copyright.
Reform tracks: to new regulations?
Faced with the challenges posed by artificial intelligence, proposals emerge to adapt the legal framework to new realities. Some provisions seem possible:
Creation of a specific status for works generated by AI
The idea would be to create a category of particular protection, which would not be exactly similar to traditional copyright. This could make it possible to give a framework to the works created by algorithms, while specifying to who belongs to this right.
License mechanisms for IA training
To respond to the concerns of creators whose works are used to train AI, it may be planned to set up a license system. Creators could be paid when their works are used to cause AI models, thus ensuring a form of compensation.
Sharing income generated
An income sharing approach could be set up for works generated by commercial purpose. In this model, the income from the creations generated by AI would be divided between the developers of the AI models, the users, and potentially the creators of the works having contributed to the training of the model. For example, if an AI model generates a commercial work derived from images, texts or protected sounds, income could include a royalty for holders of the rights of origin.
Some countries reflect or have already adopted measures. In Europe, the regulation on artificial intelligence or AI Act aims to supervise these questions, while in the United States, the Copyright Act takes into account the challenges of the works generated by AI.
Concretely, the Regulation (EU) 2024/1689 on artificial intelligencepublished in the Official Journal of the European Union on July 12, 2024, is a major step in the supervision of artificial intelligence systems in Europe. This text establishes harmonized rules aimed at regulating AI technologies in the European Union. Its provisions will come into force from August 2025.
The developers of these systems must provide detailed technical documentation. This documentation will include precise information on the operation of the models, the data used for their training, and the methods used. This will guarantee rigorous traceability and better understanding of the functioning of these technologies.
In addition, transparency obligations will be imposed. Generative AI systems must clearly inform users that the content they produce has been generated by artificial intelligence. This measure aims to avoid any confusion with human creations, thus ensuring a more ethical use of these tools.
Finally, the rules of compliance with copyright are reinforced. The data used to cause AI models must strictly respect European legislation in this area. Developers will have to ensure that their practices do not violate the rights of original creators, notably by avoiding the use of protected works without authorization. This framework seeks to prevent counterfeiting issues while protecting the interests of authors in the face of the development of AI technologies.
Conclusion
The intersection between artificial intelligence and copyright constitutes a still largely unexplored field, filled with challenges, but also opportunities. For the moment, traditional copyright laws are struggling to keep up with the frantic pace of technological advances.
The balance to be found between innovation and protection of creators is delicate. On the one hand, it is essential to support the development of technologies that promise to open up new borders. On the other, the protection of creators' rights is crucial to guarantee just remuneration and encourage artistic creation.
Ultimately, an evolution of the legal framework seems inevitable, to allow a harmonious coexistence between AI and human creators, and so that creativity, whether human or assisted by machines, continues to be valued and protected.
Maximize your Cointribne experience with our 'Read to Earn' program! For each article you read, earn points and access exclusive rewards. Sign up now and start accumulating advantages.
