Particle.news
Download on the App Store

EU States Back Ban on Non-Consensual Sexualized Deepfakes, Moving AI Act Change Forward

The push responds to the Grok chatbot’s mass production of sexualized images earlier this year.

Overview

  • EU member states’ ambassadors agreed a common Council position to outlaw AI applications that generate sexualized images of real people without consent as an amendment to the AI Act.
  • European Parliament factions reached a provisional KI‑Omnibus compromise to prohibit only non‑consensual or abusive uses, after Greens sought a broader ban that would also cover consensual creations.
  • Parliament’s Justice and Internal Market committees are set to vote next week on the compromise text, with trilogue negotiations to follow to settle the final wording.
  • The Council presidency’s draft treats two prohibitions separately by targeting systems that create non‑consensual sexualized deepfakes and systems that produce child sexual abuse material.
  • The action follows Grok’s “Spicy Mode,” which enabled rapid creation of sexualized images of women and children, triggering national probes, a EU‑mandated inquiry by Ireland’s regulator, and pledges from X to curb the feature; ambassadors also backed delaying high‑risk AI rule start dates to December 2027 and August 2028.