EU Bans Sexualized AI Deepfakes, Delays High-Risk AI Rules
The EU has agreed to ban sexualized AI deepfakes and delay high-risk AI system rules.
Why it matters: Legal and compliance teams advising clients in the EU must prepare for new AI content bans by late 2026, while having more time to adapt to complex high-risk AI rules. The moves signal evolving global standards on AI misuse and innovation, influencing corporate AI governance strategies worldwide.
- On May 7, 2026, the EU reached a deal banning non-consensual sexualized deepfakes and child abuse AI content.
- Companies must comply with the ban by December 2, 2026.
- Implementation deadlines for high-risk AI system requirements are pushed to late 2027 and 2028.
- The grace period for AI content transparency measures, such as watermarking, will now end December 2, 2026.
The European Union has reached a provisional agreement to ban AI systems that generate non-consensual sexualized deepfakes and child sexual abuse material. The regulations, finalized on May 7, 2026, prohibit marketing, deploying, or using such AI systems without reasonable safeguards.
- Compliance deadline: Companies must align their operations with the new ban by December 2, 2026.
- High-risk AI grace period: The introduction of obligations for 'high-risk' AI systems has been postponed to December 2, 2027, for stand-alone systems and August 2, 2028, if these systems are integrated into products.
- Transparency requirements: The grace period for implementing transparency measures—such as watermarking AI-generated content—has been shortened to three months, also ending December 2, 2026.
EU lawmakers positioned the agreement as both a protective and pragmatic step. Arba Kokalari, Co-rapporteur for the Internal Market and Consumer Protection committee, stated, "We now make the AI rules more workable in practice, remove overlaps and pause the high-risk requirements." Michael McNamara, Co-rapporteur for Civil Liberties, Justice and Home Affairs, added that the measures "ban nudification apps" and give the Parliament "tools to act if providers do not address AI systems that compromise fundamental rights or human dignity."
These changes reflect the EU's balancing act between safeguarding individuals from AI misuse and supporting the region's innovation ecosystem. Legal advisors should closely monitor evolving guidance and prepare compliance roadmaps ahead of the new deadlines.
Read more: European Parliament Press Release | Digital Strategy update
By the numbers:
- December 2, 2026 — Ban on non-consensual sexualized AI deepfakes takes effect
- December 2, 2027 — Stand-alone high-risk AI system rules apply
- August 2, 2028 — High-risk AI rules for integrated products take effect
- 3 months — Revised grace period for AI transparency, ending December 2, 2026
Yes, but: Enforcement and monitoring mechanisms for the new ban remain unclear.