The claim
The European Union’s technology chief, Henna Virkkunen, told Reuters on Friday that the bloc does not need new legislation to handle AI systems that slip past human control, because the existing AI Act already reaches them.
“We have our AI Act in place and the AI Act covers the whole life cycle of these models,” she said in the interview. Asked about the risks from the most capable systems, she said the issue was “a very hot topic internationally” and that Europe was “well equipped for that”. She also pushed back on the argument that the rules are already out of date, saying the legislation had “been taking into account very well already the coming developments”, while conceding that “all these capable models, of course they are posing new kinds of risks that we have to be aware of”.
The enforcement step that already happened
The concrete action behind the claim is older than the interview. The Commission sent information requests to more than 30 AI companies in late August, asking about their safety and security measures, and followed up with questions on transparency and copyright obligations. It has not named the companies. Virkkunen said she is now assessing the responses.
She also said the Commission had sent requests to Chinese AI startups, noting that the most capable models on the market come from the United States and China.
Those requests are not themselves penalties. They are the step that can open an investigation, and Reuters reports that an investigation down that road can end in fines of up to 7% of a company’s global annual turnover.
What it does not settle
Virkkunen’s position is a regulator’s assertion about the reach of a law, not a finding that the law has been applied to a rogue-agent case. No such case has been decided. The reporting ties her comments to recent incidents at OpenAI and Anthropic that raised concerns about systems acting outside their instructions, but the article gives no detail of what happened in either.
The gap between the two readings matters. A law that covers the life cycle of a model on paper still has to produce a decision about a specific product and a specific behaviour before anyone knows how wide its reach actually is. Virkkunen is arguing that the first step is unnecessary because the drafting anticipated it; the companies arguing the rules are outdated are making the opposite bet.
What to watch
Two things are checkable from here. First, whether the assessment of the August responses turns into a formal proceeding against any named provider, which is the point at which the 7% figure stops being hypothetical. Second, whether the Commission publishes anything about what it asked for and what it got, because at present neither the questions nor the answers are public.
