Enforcement powers of the EU AI Act have taken effect, with AI companies including OpenAI and Anthropic facing stricter regulation.

The European Commission has obtained new AI regulatory enforcement powers effective August 2, enabling it to conduct model assessments, restrict market access, and impose fines on companies developing general-purpose AI (GPAI) models, with U.S. AI firms including OpenAI, Anthropic, and Google as key targets. Under the implementation framework of the EU AI Act, the Commission can require companies to submit model assessments before their products enter the European market, and impose penalties for violations of up to €15 million or 3% of the company’s global annual revenue, whichever is greater. These new powers are viewed as a key measure for the EU to advance "tech sovereignty," and may further escalate tech regulatory frictions between Europe and the U.S. Earlier, the EU fined Google $1 billion under the Digital Markets Act (DMA), and former U.S. President Donald Trump once threatened to levy "massive" tariffs on the bloc. The EU AI Office noted that risks from advanced AI models are growing, necessitating strengthened oversight. Analysts point out that even if AI companies are headquartered in the U.S., they cannot evade EU regulation; non-EU AI service providers must also appoint an authorized representative within the EU to serve as a regulatory liaison. Recently, the EU has intensified communications with firms like OpenAI and Anthropic, focusing on security risks related to AI models. OpenAI confirmed it is cooperating with the EU AI Office. Tom Gordon, vice president of policy for OpenAI’s Europe, Middle East and Africa (EMEA), said the company will continue to collaborate with the European Commission and the industry ecosystem to promote the implementation of the AI Act while advancing the development of artificial intelligence in Europe.