Overview
- The European Union and the European Parliament agreed to outlaw AI tools that create sexual images, video or audio of identifiable people without consent, including content involving minors.
- Providers must add security that blocks such outputs by December 2, 2026, and single‑purpose tools for “nudifying” or generating child‑abuse material will be barred from the EU market.
- Multi‑use AI systems can remain available only if they build in reasonable safeguards that prevent this type of content from being created or spread.
- The push followed months of scandals involving Grok on X enabling nude deepfakes of adults and children and high‑profile fake images of European politicians.
- Lawmakers also delayed other duties, with watermarking of AI media due in December 2026 and high‑risk system rules starting December 2, 2027, while allowing limited personal‑data use to fix bias and carving out exemptions for products already regulated under other EU safety laws; industry group CCIA Europe said the rewrite falls short on easing compliance.