EU agrees simpler AI rules and ban on deepfake nude apps

The European Union has reached a provisional agreement to simplify parts of its AI law while introducing a ban on AI tools that create non-consensual sexual images, including “nudify” apps.

The deal was struck on Thursday by negotiators from the European Parliament and the Council of the EU as part of the bloc’s wider “Omnibus VII” plan to reduce bureaucracy and improve digital competitiveness.

Under the agreement, rules for high-risk AI systems will be delayed to give companies more time to prepare. Stand-alone high-risk AI systems will now face the rules from December 2027, while AI systems built into products will follow from August 2028.

The agreement also introduces a clear ban on AI systems that generate explicit sexual or intimate images without consent, as well as child sexual abuse material. The ban is expected to take effect from December 2027.

EU officials said the changes aim to cut administrative costs, especially for small and medium-sized businesses, while keeping protections in place. The deal also strengthens the powers of the EU’s AI Office and improves coordination between EU countries.

Belgian MEP Assita Kanko welcomed the move, calling it “a clear red line against digital sexual exploitation through AI”.

The provisional agreement must still be formally approved by both the European Parliament and EU member states before becoming law.

The debate over online safety for children is also growing in Belgium. At Politico AI and Tech Week, Flemish Media minister Cieltje Van Achter warned against introducing strict ID-based age checks for social media users. She said asking the “whole population” to identify themselves online would fundamentally change how social media works and could raise privacy concerns.

Van Achter confirmed last year that Flanders would not introduce a blanket social media ban for children under 16. Instead, her government wants stronger enforcement of existing age limits, tighter rules on addictive platform features and better protection against harmful content, disinformation and AI-generated manipulation.

^{© An AI girl generator Stefani REYNOLDS / AFP}