The European Union’s AI regulators gained new enforcement powers on 2 August, allowing them to investigate artificial intelligence providers, evaluate models, restrict access to systems deemed dangerous and impose fines as the bloc steps up implementation of its landmark AI Act.
The European Commission said new transparency obligations have simultaneously come into force, requiring certain AI-generated or manipulated content, including deepfakes, to be clearly labelled and machine-readable, while users must be informed when interacting with AI systems such as chatbots or avatars.
The rules are designed to help people distinguish AI-generated content from authentic material and reduce risks including misinformation, fraud and impersonation.
The Commission said responsibility for enforcement is shared between its AI Office, national market surveillance authorities and the European Data Protection Supervisor. The AI Office will oversee general-purpose AI models, with powers to request information, demand access to models for evaluation, require providers to address identified risks and, where necessary, restrict systems from the EU market.
Companies that breach obligations can face fines of up to €15 million or 3 per cent of global annual turnover, while prohibited AI practices can attract penalties of up to €35 million or 7 per cent of worldwide turnover.
Henna Virkkunen, executive vice-president for tech sovereignty, security and democracy at the European Commission, said: “Harms can occur if AI is not properly designed and used and the most advanced models create risks on an entirely new scale.”
The new powers increase regulatory scrutiny for major AI developers including OpenAI, Anthropic and Google, regardless of where they are based. Elisabetta Righini, partner at Sidley Austin, told CNBC: “A US address does not put a lab outside the EU regulator’s reach,” adding that non-EU providers must appoint an EU-based authorised representative. She said exposure to regulatory penalties was “real”, noting that refusing information requests or obstructing model evaluations could itself result in fines.
OpenAI’s Tom Gordon, vice-president for EMEA policy, told CNBC: “We’ve collaborated closely with the European Commission and the wider ecosystem on implementing the AI Act, including its Codes of Practice, and will continue working together to help Europe realise the benefits of the Intelligence Age.”
The Commission’s expanded powers follow recent AI safety incidents involving frontier models and come as the AI Office increases its regulatory capacity, with plans to recruit 40 additional contract staff through 2027 to strengthen compliance and AI safety work.
