France: Senate Mandates Transparency for AIs on Use of Works
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
The French Senate has taken a decisive step by adopting, on Wednesday, April 8, a bill that disrupts the legal framework surrounding the use of cultural works by artificial intelligences. This innovative text establishes a presumption of use of protected works by AIs, shifting the burden of proof onto the developers of these technologies. From now on, it will be the tech companies that must demonstrate they have not exploited cultural repositories to train their models.
This legislative initiative, led by Laure Darcos and her fellow senators, aims to rebalance the relationships between Silicon Valley and the creative world. Until now, artists, whether photographers or writers, had to prove that their works had been used to feed algorithms, a task made nearly impossible by the opacity of training databases, often referred to as "black boxes."
A Transparency Challenge for Tech Giants
With this new law, companies like OpenAI or Anthropic will need to thoroughly revise the architecture of their training logs to comply with future judicial requirements. Laure Darcos clarified that this obligation will apply not only to model providers but also to the AI systems that use them. In the event of a dispute, an artist can now ask a judge to require concrete evidence from the tech giants to dismiss the presumption of theft.
Laurent Lafon emphasized the importance of this measure to ensure the effectiveness of the European right to opt-out, which until now has remained largely theoretical due to a lack of control mechanisms. The modification of the Intellectual Property Code will require industry players to accurately document every data source used. Those who cannot justify the origin of their billions of parameters risk automatic penalties. This law will even apply to ongoing legal proceedings, which could expedite the resolution of existing disputes between media publishers and research laboratories.
A Unique Legislative Advancement
France stands out on the international stage with this bold legislative measure, although the text still needs to be validated by the National Assembly. This initiative could conflict with the interests of French startups, such as Mistral AI, which fear that overly strict traceability obligations may hinder their ability to innovate in the face of American giants.
Despite these concerns, Laure Darcos maintains that this legislation is crucial for protecting the cultural sovereignty of the country. The Council of State confirmed on March 19 that this law complies with the Constitution and the legal framework of the European Union. If it is ultimately adopted, France will become the first country in the world to implement such legislation. However, debates in the National Assembly are expected to be complex, as deputies will need to find a balance between protecting authors and maintaining the competitiveness of the French tech industry.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.