A Bill in France Would Force AI Firms to Prove They Didn’t Use Copyrighted Content to Train Their Tech. The Country’s Music Biz Is Urging Lawmakers to Pass It

Why It Matters

The proposal forces AI companies to demonstrate compliance, potentially reshaping how generative AI is trained and protecting creators’ rights across Europe. It could become a template for other jurisdictions grappling with AI‑driven copyright infringement.

Key Takeaways

• 81 French cultural groups push AI copyright presumption bill
• Bill reverses burden of proof to AI providers
• Senate passed Darcos law unanimously; Assembly yet to schedule
• EU and UK consider similar measures amid rising AI lawsuits

Pulse Analysis

Generative AI’s rapid growth has sparked a global debate over the legality of training models on copyrighted material. In the United States, major music publishers have filed multi‑billion‑dollar lawsuits alleging unauthorized use of songs, while the UK recently reversed a policy that would have eased AI access to music. These actions reflect mounting pressure from creators who argue that AI firms are profiting from content without proper compensation, prompting lawmakers worldwide to reconsider existing copyright frameworks.

France is taking a decisive step with the Darcos bill, a legislative proposal that inserts a new article into the French Intellectual Property Code. The measure establishes a rebuttable presumption that any AI system has used protected works unless the provider can demonstrate otherwise, effectively flipping the traditional burden of proof. The bill, championed by Senator Laure Darcos and unanimously approved by the Senate, now awaits scheduling by the National Assembly. A coalition of 81 cultural, film, publishing and press organisations—including SACEM, SNEP and major screen‑writer societies—has rallied behind the proposal, arguing it restores balance between creative sectors and tech companies.

If enacted, the French presumption could set a precedent for the European Union, where legislators are already discussing similar safeguards in the EU AI Act and recent Parliament resolutions. By compelling AI developers to verify the provenance of their training data, the law could drive the creation of transparent licensing models and new revenue streams for rights holders. Conversely, it may increase compliance costs for AI firms and influence where they locate research operations, potentially reshaping the competitive landscape of generative AI across the continent.