EU member states have reached a landmark agreement to ban AI systems that produce sexually explicit images without the consent of the individuals depicted. The decision,

confirmed Friday by a spokesperson for the Cypriot Presidency of the Council of the EU, marks a decisive step in regulating high-risk AI applications.

The move follows earlier controversies surrounding Grok, the AI assistant of Elon Musk’s social media platform X. In January, the chatbot generated millions of non-consensual nude images of real people—including minors—sparking global outrage. Despite X’s assurances of a “zero-tolerance” policy toward sexualized deepfakes and measures to curb their creation, reports indicated that the AI’s undressing feature remained accessible. The European Commission subsequently launched an investigation into the platform.

Belgium has emerged as a key supporter of the ban. Vanessa Matz, Belgium’s federal minister for Digitalisation, said the country typically abstains from EU AI proposals due to internal disagreements. This time, however, Belgium joined other nations advocating stronger protections for citizens, particularly girls and women, who are most often targeted by such abuses.

“It is not just about individual scandals such as Grok. It is about how much power we are prepared to give AI to humiliate people,” said MEP Sergey Lagodinsky, a proponent of the legislation, speaking to The Straits Times.

The EU Parliament is preparing a parallel amendment, with a vote scheduled for 18 March. Once passed, member states and the Parliament must finalize joint wording. Rules governing high-risk AI applications are expected to take effect in August 2026 and August 2027, unless formally delayed by law.

This ban represents a broader effort by the European Commission to simplify AI legislation and address emerging threats posed by AI technology while safeguarding individual privacy and dignity.