The European Union has reached a provisional agreement to ban AI-powered "nudify" apps — tools that generate fake sexually explicit images of real people without their consent — while also simplifying broader AI regulations for businesses. The deal, struck after nine hours of overnight negotiations between EU member states and European Parliament lawmakers, marks a significant step in addressing what advocates have called a growing crisis of digitally facilitated sexual violence.

Nudify applications use artificial intelligence to remove clothing from existing photographs or generate entirely fabricated nude or sexually explicit images of real individuals. The results are often highly realistic and are frequently produced and shared without the knowledge or consent of the person depicted. The issue gained particular prominence in Europe following controversy over Elon Musk's xAI chatbot Grok, which was used to generate such images of women and, in one widely reported instance, produced sexualised depictions of teenage girls before the feature was restricted. The new ban will explicitly cover the creation of AI-generated child sexual abuse material as well. Under the agreement, mandatory watermarking of AI-generated content will also take effect from 2 December 2026, when the ban is due to enter force, enforced primarily by the EU's AI Office, which was established two years ago.

The deal also makes significant concessions to business interests. Rules governing high-risk AI systems — including those involving biometrics, critical infrastructure, and law enforcement — will be delayed from an original August 2025 deadline to December 2027. Machinery will be excluded from the AI Act's scope entirely, as it is already regulated under separate sectoral rules. These changes come in response to sustained pressure from companies that argued overlapping regulations and administrative burdens were hampering Europe's ability to compete with rivals in the United States and Asia. Cyprus, which currently holds the rotating EU Council presidency, described the agreement as significantly reducing recurring administrative costs for businesses.

The provisional deal still requires formal endorsement by EU governments and the full European Parliament, a step that is generally considered procedural. European lawmakers welcomed the outcome with cautious optimism. "Every day, people — especially women and children — are victimised by deepnude tools that digitally strip, humiliate and expose them to blackmail and abuse," said Belgian MEP Kathleen Van Brempt. Dutch MEP Kim van Sparrentak added that by the end of the year, nudifier apps would effectively be removed from the EU market.

Why this matters: deepfake sexual imagery is widely recognised as a form of gender-based violence, with women and girls disproportionately targeted. Until now, the AI models used to create such content operated in a legal grey zone even where the images themselves were illegal to distribute. The EU's agreement closes that gap — though critics note that robust enforcement, rapid image takedown mechanisms, and accountability for the platforms hosting such content remain unresolved challenges.