The European Commission has activated new enforcement powers under the EU AI Act that allow it to scrutinize major artificial intelligence model providers, including companies such as Anthropic, OpenAI and Google. Under the Act’s enforcement framework, the Commission can request access to and inspection of AI models, including demanding model evaluations or related testing information. The Commission also has the authority to restrict access to the EU market for AI systems or models that it determines do not meet regulatory requirements.
In addition to inspection and potential market-access limits, the Commission can impose administrative penalties. Reported figures indicate fines can reach up to 15 million euros or 3% of a provider’s total annual turnover, depending on the specific circumstances and applicable provisions.
The newly active powers are part of the EU’s broader effort to ensure compliance by general-purpose AI providers and to standardize enforcement through the AI Act as oversight ramps up. The exact scope of actions against any individual firm depends on the outcome of the Commission’s review and any subsequent compliance determinations.
