detect·deepfakesby Resemble AI
Deepfake law · France

Deepfake Law in France

France combines the EU AI Act with domestic provisions from SREN Law (2024) criminalizing deepfake sharing and manipulated sexual content. Civil remedies via Article 9 of the Civil Code are also available.

Status
enacted
Jurisdiction
France
Effective
May 2024
Statute
EU AI Act + SREN Law 2024 (Article 226-8 Penal Code)
AI Act transparencyNon-consensual sexual deepfakesImage-based abuseDefamation
Updated Apr 16, 2026 · 2 min read

France's deepfake framework pairs the EU AI Act with the 2024 SREN Law (Loi visant à sécuriser et à réguler l'espace numérique), which amended Article 226-8 of the Penal Code to criminalize sharing deepfakes of real persons without consent.

Key provisions

SREN Law 2024 — Penal Code Article 226-8. Publishing a deepfake depicting a real person without their consent is punishable by up to one year imprisonment and €15,000 fine (increased to two years and €45,000 for sexual content). The offense requires that the content could reasonably be mistaken for authentic media.

EU AI Act. Directly applicable; Article 50 transparency obligations enforced through the CNIL and ARCOM (national supervisory authorities).

Civil Code Article 9. France's long-standing right to private life and image doctrine. Victims can seek damages and injunctions independently of criminal charges. French courts award damages in the €5,000–€50,000 range for image-rights violations, with higher awards in aggravated cases.

Electoral Code. Deepfakes in the context of an election are covered by existing provisions against misinformation that distorts electoral outcomes.

ARCOM (media regulator)

ARCOM, France's audiovisual and digital communications regulator, has new powers under the SREN Law:

  • Can order removal of deepfake content from platforms.
  • Can impose fines up to 1% of global turnover on non-compliant platforms.
  • Publishes guidance on AI-content labeling compatible with EU AI Act Article 50.

Practical implications

For organizations operating in France:

  • Platforms: ARCOM takedown orders are binding. Compliance infrastructure must include deepfake detection and rapid removal.
  • Creators of deepfake content: publishing without consent is criminal. Consent verification procedures are essential for any legitimate synthetic-media use case.
  • Enterprises: combined civil (Article 9) + criminal (Article 226-8) + AI Act exposure makes France among the higher-risk jurisdictions for AI-content operations.

Sources