EU Council Proposes AI Nudification Ban in Updated AI Act

Summary

The EU Council proposed AI Act amendments banning AI-generated non-consensual sexual imagery, revising high-risk AI deadlines and registration rules, and restoring strict necessity for sensitive data used in bias detection.

Key points

The Council’s proposal would ban AI practices that generate non-consensual sexual or intimate content and child sexual abuse material.
The proposed deadlines for high-risk AI rules are 2 December 2027 for stand-alone systems and 2 August 2028 for systems embedded in products.
Providers would have to register high-risk AI systems in the EU database even when claiming an exemption.
The proposal restores a strict-necessity standard for using special-category personal data in bias detection and correction.
National AI regulatory sandboxes would be due by December 2027 under the proposal.
The article says the proposal follows reports that Grok generated millions of non-consensual intimate images and a European Commission investigation.

Article Details

Event Type: The European Council proposed amendments to the EU AI Act, including a ban on AI practices that generate non-consensual sexual or intimate content and child sexual abuse material.
Impact: If adopted, the amendments would strengthen safeguards and oversight for AI systems, restore mandatory registration of high-risk systems even when providers claim exemptions, and require a strict-necessity basis for using sensitive personal data in bias detection. They would also delay certain high-risk AI compliance deadlines. The proposal follows a reported Grok incident in which the chatbot generated millions of non-consensual intimate images that spread online; the European Commission launched a formal investigation into the platform and its AI features.