The European Parliament has taken a significant step towards banning artificial intelligence systems that generate non-consensual sexual images, following a committee vote on Wednesday advancing new rules under the bloc’s AI Omnibus legislation.
The parliament’s civil liberties committee has approved an amendment that would outlaw AI tools capable of producing realistic depictions of sexually explicit content involving identifiable individuals without their consent. The measure aligns with a similar position already agreed by EU member states, increasing the likelihood that the ban will be enacted later this year.
The proposed rules specifically target AI systems rather than individual users, reflecting lawmakers’ concerns about the rapid advancement of generative models. Developers would be required to demonstrate that safeguards are in place to prevent the creation of such content, although questions remain over how consent could be verified in practice.
The initiative follows widespread backlash earlier this year after xAI’s Grok chatbot, available via X, was used to generate large volumes of sexualised images of women and children. Sergey Lagodinsky, a member of the European Parliament, told France24 “it’s about how much power we are willing to give AI to degrade people”.
The controversy prompted xAI to introduce restrictions on the tool, while regulators in both the EU and the UK launched investigations into whether the incident breached online safety and content moderation laws.
RTÉ reported that lawmakers framed the ban as a clear boundary for AI use. Independent MEP Michael McNamara said “AI must never be used to humiliate, exploit or endanger people,” adding that the parliament was “sending a clear message that this is a priority”.
The proposed prohibition builds on existing EU legislation addressing non-consensual intimate imagery and child sexual abuse material, including provisions under the Digital Services Act and a 2024 directive on violence against women. It marks the first attempt to directly regulate the capabilities of AI systems in this area.
Alongside the ban, lawmakers are revising the timeline for implementing high-risk AI rules. Bloomberg reported that key provisions originally due in August 2026 could be delayed until December 2027 and August 2028, giving companies more time to comply and regulators more scope to develop technical standards.
Negotiations between the European Parliament and member states will determine the final text of the legislation before it can be formally adopted.
