French press industry takes legal action against AI training databases
French news industry groups representing 800 media outlets filed cease-and-desist letters against public web databases Common Crawl, C4 and Oscar, alleging they distribute copyrighted press articles without authorization to train generative AI systems. The action claims these databases supply massive amounts of protected content to services like ChatGPT.
What happened
On September 1, 2025, two professional organizations representing 800 French news titles, Apig and SEPM, announced a coordinated legal action against public web databases Common Crawl, C4 and Oscar.
The organizations issued cease-and-desist letters accusing the databases of distributing millions of copyrighted news articles without authorization to train generative AI services like ChatGPT.
The presence of French press content in these databases was documented by a court officer (huissier), according to the organizations.
Reported impact
Apig and SEPM stated that generative AI services massively source press content from these databases, which reproduce and distribute copyrighted and neighbouring-rights-protected articles without authorization or access restrictions.
Index analysis
Common Crawl is a public web library maintained by robots that crawl billions of pages and publish the content in free access.
C4 and Oscar are derivatives of Common Crawl used for training generative language models.
Governance controls
Pierre Petillault, director general of Apig, stated that the legal framework is imperfect because copyright and neighbouring rights exceptions do not apply in this case.
Petillault noted tension between innovation interests and intellectual property protection, with public authorities caught between funding journalism and promoting AI development in Europe.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.