EU Strikes Deal to Ban AI-Generated Sexualized Deepfakes — Delays High-Risk AI Rules to 2027
Summary
The European Union has reached a landmark agreement to explicitly ban AI systems that generate sexualized deepfakes, marking the first time EU legislation directly targets so-called “nudifier” applications. The deal, negotiated between the European Parliament and EU member states, will be incorporated as amendments to the bloc’s comprehensive AI Act adopted in 2024.
The move comes after widespread outrage earlier this year over nonconsensual nude images produced by Elon Musk’s chatbot Grok. MEP Michael McNamara stated that the EU has “drawn a red line” and that “AI must never be used to humiliate, exploit or endanger people.”
Separately, EU negotiators agreed to delay the implementation of high-risk AI rules — governing models deemed potentially dangerous to safety, health, or fundamental rights — from August 2026 to December 2027 for standalone systems and August 2028 for embedded AI tools. The EU executive framed the delay as a measure to support business innovation while still advancing safe AI development through other provisions.
Source
Courthouse News — EU strikes deal to ban sexualized AI deepfakes (via AFP)
Commentary
The deepfake ban is a clear, necessary step — the proliferation of AI nudification tools has caused real harm to real people, and legislative action was overdue. That said, enforcement will be the real test. These tools are trivially deployable via open-source models, and a ban doesn’t make the capability disappear.
The more interesting signal is the delay of high-risk AI rules. The EU is effectively admitting it set compliance timelines before the technology — and the business landscape around it — had stabilized. Whether you read this as pragmatic flexibility or regulatory retreat depends on your perspective, but it shows the EU is learning that AI regulation requires iteration, not just ambition. The mention of Anthropic’s restricted Mythos model and the 30 MEPs calling for cybersecurity rule revisions suggests Brussels is increasingly anxious about the gap between AI capabilities and their ability to oversee them.
