The European Commission has started enforcing the EU AI Act's rules for general-purpose AI models, granting it new powers to inspect models, restrict market access, and impose fines up to €15 million or 3% of annual turnover. U.S. companies like OpenAI, Anthropic, and Google are now subject to these regulations.
TL;DR
- EU AI Act enforcement began August 2, targeting general-purpose AI models with fines up to €15M or 3% of annual turnover.
- OpenAI, Anthropic, and Google are among the U.S. companies subject to new EU regulations.
- Companies must document model information, establish copyright policies, and meet transparency requirements.
What happened
The European Commission's AI Office began enforcing the EU AI Act's rules for general-purpose AI models on August 2. These rules apply to all companies, regardless of headquarters location, that make such models available in the EU, according to CNBC.
Under the AI Act, providers must document specific information about their models and share it with authorities or downstream providers. They must also establish a copyright policy and publish a summary of the data used for training their models.
Providers of the most advanced models, defined as those posing systemic risks, face additional requirements to address potential large-scale harms, including cybersecurity threats and risks to fundamental rights, the Commission said.
Why it matters
The enforcement regime covers prohibited AI practices, such as systems that manipulate people or exploit vulnerabilities. It also introduces new transparency requirements, mandating that chatbots and other interactive AI systems disclose their AI nature to users.
Companies like OpenAI, Anthropic, and Google must comply with these regulations or face fines up to €15 million or 3% of their annual turnover, whichever is higher. This includes potential fines for refusing information requests or blocking model evaluations, according to Elisabetta Righini, partner at law firm Sidley Austin.
The EU has also opened discussions with OpenAI and Anthropic following cyber attacks linked to their models. This enforcement action comes after months of tension between the EU and U.S. AI labs over model access.
Key facts
- Enforcement of the EU AI Act began on August 2, 2026.
- Fines for non-compliance can reach up to €15 million or 3% of annual turnover.
- OpenAI, Anthropic, and Google are among the U.S. companies subject to the new regulations.
- Providers must document model information, establish copyright policies, and publish training data summaries.
- Advanced models face additional requirements to address systemic risks.
- Chatbots and interactive AI systems must disclose their AI nature to users.
- Companies outside the EU must designate a local representative for regulatory contact.
- The EU has opened discussions with OpenAI and Anthropic following linked cyber attacks.
Context
The EU AI Act represents a significant step in regulating AI technologies, aiming to ensure safety, transparency, and accountability. This enforcement action underscores the EU's commitment to mitigating potential harms posed by advanced AI models.
The regulations are part of a broader effort to address the risks associated with AI, including cybersecurity threats and the protection of fundamental rights. The EU's actions highlight the growing importance of international cooperation in AI governance.
For AI developers, startups, and investors, these regulations present both challenges and opportunities. Compliance with the EU AI Act may require significant investments in documentation, transparency, and cybersecurity measures, but it also underscores the importance of ethical AI development and the potential for market access in the EU.
