Since August 2^nd, 2026, the Regulation (EU) 2024/1689 of the European Parliament and of the Council of June 13^th, 2024, knows as AI Act, has reached a significant milestone with the entry into force of its provisions.

This date marks the aplicability of obligations pertaining to high-risk and limited-risk systems, categories we will discuss more hereafter, and in particular the imposition of transparency requirements on suppliers, with the aim of promoting trustworthy AI for all users.

These information and transparency obligations raise the question of how they relate to preexisting rules under French copyright law, particularly the right of authorship granted to the authors of an original work. Given that domestic law is still insufficient in this field, the AI Act appears to be the first step toward establishing a regulatory framework for the author’s moral rights.

Preexisting recommendations regarding the protection of audiovisual works in the face of AI

On May 19^th, 2024, the French CNC published an initial study titled « Artificial intelligence observatory », which examined the uses and impacts of AI among professionals in the French audiovisual industry. The study already revealed that more than 45% of the surveyed screenwriters, directors and cinematographers used AI, as did more than 69% of producers, for purposes such as research, translation, inspiration, or testing new ideas.

These percentages have risen significantly over the years, as evidenced by the third study published in June 2026: more than 58% of surveyed screenwriters, 80% of producers, 60% of directors, and nearly 64% of cinematographers now use AI.

Given this trend toward increasingly widespread use of AI (generative AI, or GAI), the question of regulatory oversight has inevitably arisen. In practice, it appears that contractual provisions related to AI are the subject of intense negotiations between producers and talents, whose interests at stake sometimes seem to be at odds with one another. A producer will seek to minimize the use of AI in the context of promoting or post-producing the work, while also seeking to protect themselves from an author’s unrestricted use of AI, which could impact the traceability of rights and the quality of their work. The author, for their part, will seek to protect and control their image and their work from unreasonable use of AI by their producer, while many issues regarding image control will be a source of concern for acting talents.

Although lawmakers have not fully addressed these issues, recommendations have been implemented by organizations that manage and support authors for several years now.

At the initiative of the French SACD and representatives of producers, as of October 23^th, 2024, five key principles were established to guide the drafting of clauses governing the use of GAI in contracts for authors of audiovisual works. These principles are based, notably, on full transparency and the requirement for producers and authors to inform one another regarding the use of GAI by either party, as well as on the prohibition against requiring authors to use such a system as part of their work.

More recently, in a press release dated November 17^th, 2025, organizations in the documentary field established best practices to ensure, in particular, compliance with copyright law and the responsible and transparent use of GAI systems. Standard clauses were thus established stipulating, once again, that the producer is prohibited from requiring the author to use GAI in any way as part of their work, and that the producer must inform the other party and/or seek their authorization in the even that such use is claimed. 

While these recommendations have only limited binding force, their significance is underscored by the recent entry into force of numerous provisions of the AI Act that promote greater oversight of AI use by talents and producers.

AI Act provisions newly in effect

Article 3 of the AI Act defines AI systems (AIS) as « a machine-based system that is designed to operate with varying levels of autonomy and that may exhibit adaptiveness after deployment, and that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments ».

The AI Act distinguishes between different categories of AIS, classified by severity and the strictness of regulatory oversight: certain systems and related practices are outright prohibited, such as real-time biometric identification or social scoring; so-called « high-risk » AIS, such as the management and operation of critical infrastructure or education and vocational training; AIS considered « limited-risk », notably chatbots and content-generation systems; and finally, AIS deemed to pose a minimal risk, such as the use of AI in video games.

GAI systems, which raise questions regarding copyright law, therefore fall into the third category. The AI Act sets out general obligations for suppliers of these GAI systems, which are not particularly restrictive, such as drafting and updating technical documentation and information available to users, or implementing a usage policy that complies with the various copyright laws of EU Members. 

Most importantly, Article 50 of the AI Act, which took effect on August 2^nd, 2026, requires suppliers and operators of AIS to comply with transparency obligations regarding any use or reliance on such a system, so that consumers are clearly and explicitly informed at all times that they are interacting with or viewing content generated, in whole or in part, by AI.

The second paragraph of the aforementioned Article 50 requires the labeling of content (audio, images, video or text) that has been generated, in whole or in part, by AI. Of course, this labeling requirement applies to practices not expressly prohibited by the AI Act; such practices are prohibited if their purpose is to deceive or manipulate the user. Icons for labeling AI-generated content have therefore been created to facilitate the reporting required by the transparency obligation. The fourth paragraph of that article specifies an enhanced transparency obligation in the case of deepfakes.

Regarding liability, the AI Act provides for penalties against suppliers and operators of AIS that fail to comply with the provisions set forth therein, particularly the transparency rules. Fines of up to 15 million euros and 3 percent of a company’s global annual revenue may be imposed on them.

In order to ensure compliance with the AI Act regulations, the European AI Office and the European Data Protection Supervisor were established.
In France, with regard to compliance with transparency obligations, the authorities designated as competent are the French DGCCRF, CNIL and Arcom.

The recent entry into force of the transparency and information requirements set forth in the AI Act aligns with the French guidelines that already existed in that area. These new provisions will undoubtedly make it easier to monitor the implementation of contractual agreements between the various professionals in audiovisual production. Given the complexity of applying the GAI within the context of copyright law, it is advisable to consult with a specialized attorney to ensure compliance with the new provisions and to manage this new issue consistently through contractual agreements.