EU lawmakers move to ban nudify apps after Grok goes off the rails

The European Union is finally getting serious about AI-generated nude images, and Elon Musk’s Grok chatbot is the poster child for why.

This week, the European Parliament’s Internal Market and Civil Liberties committees voted 101–9 (with 8 abstentions) to simplify the AI Act and explicitly ban “AI ‘nudifier’ systems.” The joint press release didn’t name Grok directly, but it didn’t have to.

Earlier this year, the European Commission admitted that the current AI Act doesn’t actually prohibit “AI systems that generate child sexual abuse material (CSAM) or sexually explicit deepfake nudes.” That’s a pretty glaring hole in a regulation that was supposed to be the world’s gold standard for AI safety.

Grok made that hole impossible to ignore. Musk’s chatbot became a prime example of what happens when an AI platform fails to block outputs that sexualize real people. The platform generated sexually explicit deepfake images, including of children, and then Musk tried to blame the users. That tactic might work in some jurisdictions, but EU lawmakers clearly aren’t buying it.

The vote signals that Parliament members were already working on amendments to close those loopholes. The Commission’s finding seems to have lit a fire under them. 101 votes in favor versus 9 against isn’t a close call — it’s a mandate.

What’s interesting here is the framing. They’re not just banning deepfake porn generically. They’re targeting “nudifier” systems specifically — apps and tools designed to strip clothing from images. That’s a narrower target than I expected, but it’s also the most egregious category. These services have been proliferating on the web for years, often targeting women and girls without consent.

The real question is enforcement. The AI Act has extraterritorial reach, meaning any company operating in the EU market has to comply. But we’ve seen time and again that tech companies — especially the ones run by free-speech absolutists — will push boundaries and wait for the fines to roll in.

Grok’s defenders will argue that any image generator can be misused. That’s technically true, but it misses the point. The difference between a general-purpose tool that occasionally gets abused and a system that actively strips clothing from photos is the difference between a knife and a guillotine. One has legitimate uses; the other is designed for a single purpose.

I’m curious to see how this plays out with the broader AI Act simplification. The committees voted to “simplify” the law, which could mean anything from streamlining compliance procedures to weakening protections. The fact that they’re simultaneously adding new bans suggests they’re trying to thread a needle — make the law easier for good actors to follow while closing doors for bad ones.

Musk’s blaming of users for Grok’s failures was always a weak defense. If your AI platform can be prompted to generate CSAM, the problem isn’t just the prompt — it’s the system that didn’t filter it. EU lawmakers seem to agree, and they’re moving fast to make sure the law catches up with the technology.