SACEM on Alert: Copyright Law at Risk from AI

SACEM Calls for Action to Save Copyright Law Against AI

This Tuesday morning, SACEM published an urgent collective statement, urging lawmakers to immediately place the law on the presumption of use of works by artificial intelligence on the agenda. This initiative aims to mobilize not only its members but also the lawmakers themselves.

The proposed law, unanimously adopted by the Senate on April 8, introduces a presumption of use of cultural content by AI providers. However, it is currently stalled in the National Assembly. Assigned to the Cultural Affairs Committee as of April 9, it has not progressed since. SACEM, which manages the copyright of authors, composers, and music publishers, refuses to wait any longer and calls for immediate mobilization from creators.

The French law, which would presume AI providers guilty of using works without authorization, risks never being voted on. Can France effectively legislate on the dangers of AI for content creation? In December 2025, Senator Laure Darcos submitted a short but radical text, consisting of a single article. This text raises a crucial question: who must prove what? Until now, it was the creator's responsibility to prove that their works had been used to train an AI model, a nearly impossible task due to the inaccessibility of training data for models. The proposed law aims to reverse this situation: now, it would be the technology company’s responsibility to prove that it did not use anything.

AI companies have never opened their books. In other words, the training databases of large generative models are true vaults. The data used to train their models, such as texts, images, and music, remains completely secret. An author cannot know if their novel, photo, or song is included. It is precisely this wall of opacity that the law seeks to circumvent. Without needing to provide proof, a creator will now be able to take legal action.

On April 8, 2026, the Senate unanimously adopted this text, with support from the right, left, and center, without a single dissenting vote. A rare consensus in a chamber often divided by political fractures. The text then followed the classic legislative path. Transmitted the next day to the National Assembly, it was assigned to the Cultural Affairs Committee, responsible for examining it before a potential vote in a plenary session. But since April 9, nothing has happened.

SACEM Alerts Its Members on the Urgency of the Vote in the National Assembly

Faced with the inaction of the National Assembly, SACEM has decided to take action. The organization that collects and redistributes the rights of authors, composers, and publishers has contacted its members to invite them to sign a collective statement, published this Tuesday, April 28, and directly addressed to lawmakers. The goal is to create sufficient public pressure for lawmakers to place the text on the agenda before it is too late.

In this text, the signatories candidly describe what is happening. Their works are "sucked in and digested by AI systems that learn from them, learn about them and us, without our consent, without any compensation or remuneration," the letter states. A conviction that creators now believe is supported by facts: some tech giants have acknowledged using protected works, and in the United States, several of them have already paid millions of dollars to settle copyright infringement lawsuits.

SACEM emphasizes its concern. It states, "the adoption of this text hangs by a thread." Two scenarios particularly worry the management company:

• If lawmakers do not quickly place the text on the National Assembly's agenda, it may never be put to a vote.
• This proposed law was adopted by the Senate in a specific version. If a lawmaker modifies it, "even just once," the text will have to go back to the Senate for a new vote, a process that could effectively seal the text's fate.

France: Pioneer or Hostage to Lobbying by AI Giants?

While creators sign the statement, major AI platforms are advancing their pieces, with lobbyists making numerous contacts with lawmakers to convince them that this law would harm technological innovation. SACEM anticipates this argument and counters it directly in the text, explaining that "respect for intellectual property is not a hindrance to innovation; rather, it is the condition for its legal security and legitimacy." In other words, an AI that develops by plundering creators' works builds its success on shaky legal ground.

The law does not please everyone in France either. French start-ups like Mistral AI fear that the obligation to precisely document the origin of each piece of data used to train their models will significantly slow them down, while their much more powerful American competitors would more easily absorb these constraints. Lawmakers thus find themselves facing an uncomfortable dilemma: to protect French creators or not to hinder French AI companies in a global race they are far from winning.

If lawmakers vote for this text, France will become the first country in the world to require AI providers to prove that they have not used creators' works without authorization. A strong symbol for the nation that invented copyright in the 18th century and could, three centuries later, redefine the rules in the age of artificial intelligence. Members, however, have no time to waste on symbols. "Time is of the essence," concludes SACEM.