The European Union's Artificial Intelligence Act has moved from statute book to enforcement, and the companies building the most powerful AI systems are the first ones in the frame.
According to CNBC, Anthropic and OpenAI are among the firms facing new scrutiny as the European Commission's enforcement powers take effect. Those powers are substantial: CNBC reports the Commission can demand to inspect AI models, restrict access to the European market, and fine providers up to 15 million euros or 3% of turnover.
That last detail is the one to watch. A percentage-of-turnover penalty scales with company size, which means the ceiling is not a fixed cost to be budgeted around — it grows with the business. The power to inspect models is arguably just as significant, since it pushes regulators past reviewing paperwork and toward examining the systems themselves.
Other outlets frame the same moment in similar terms. Podnews reports that the EU Artificial Intelligence Act becomes law. The Capitol Forum describes the EU as gaining power to regulate generative AI and Big Tech. Benzinga characterizes it as a new EU AI crackdown as regulators gain enforcement power.
Neither Anthropic nor OpenAI is described in these reports as having been penalized. The shift is that a supervisory regime now exists with the authority to look inside their products and to block them from a market of roughly 450 million people.
It matters because the EU is setting the first binding rulebook for general-purpose AI, and whatever compliance looks like in Brussels will likely shape how these systems are built and shipped everywhere else.