EU AI Act: Rules on AI Models Take Effect Today

The EU AI Act's provisions on general-purpose AI models become enforceable on August 2, 2026, making Brussels the world's most prominent AI regulator.

As of today, August 2, 2026, the European Union's AI Act rules on general-purpose AI models become enforceable, cementing the European Commission's role as the world's most prominent AI regulator.

The AI Act, the first comprehensive law regulating artificial intelligence, now requires developers of large language models and other general-purpose AI systems to disclose how their models were built, reveal any copyright-protected content used for training, and provide enough information for downstream users to understand the model's capabilities.

Transparency takes center stage. From today, chatbots and interactive AI systems must inform users they are dealing with AI, not a human. Deepfakes — AI-generated or altered images, videos, and audio — must be clearly labelled. AI-generated content must also carry machine-readable watermarks. More than 180 organisations have already signed the Commission's Code of Practice on transparency of AI-generated content.

Frontier models face additional scrutiny. Companies developing the most powerful AI systems must now identify and mitigate systemic risks to society — a provision that could cover models with advanced cyber capabilities or those raising existential safety concerns.

Industry reactions are split. OpenAI's VP of EMEA Policy Tom Duff Gordon told Euronews the company has "collaborated closely with the European Commission" on implementation. Most leading Western AI labs signed the voluntary code of practice — Meta is the notable holdout.

Geopolitical friction looms. MEP Michael McNamara warned Euronews that "the danger is that the current US administration treats this as an attack on US commercial interests," citing past clashes over the EU's Digital Markets Act enforcement against American tech firms.

What it means for users. In practice, Europeans may see some advanced AI models launch in the EU a few weeks later than in other markets as companies complete compliance checks. The trade-off: AI models available in the EU will, in theory, have passed safety vetting that other jurisdictions don't yet require.

The European AI Office, working alongside national authorities, will lead enforcement — though its resources are knowingly limited against some of the world's richest companies. The EU is effectively setting the global benchmark for AI governance, with other jurisdictions still taking a wait-and-see approach.

Sources: Euronews, European Commission