EU AI Act: general-purpose AI rules apply from 2 August 2025
In one sentence From 2 August 2025 the AI Act obligations for providers of general-purpose AI (GPAI) models apply. The voluntary Code of Practice, published on 10 July, is signed by most major labs but not by Meta. The Commission's enforcement powers start on 2 August 2026.
The EU AI Act, in force since 2024, applies in stages. On 2 August 2025 the obligations for providers of general-purpose AI (GPAI) models kick in, meaning the big models such as GPT, Claude, Gemini or Llama.
In practice anyone placing a new model of this kind on the EU market must keep technical documentation, give information to those integrating it into their products, have a copyright-compliance policy and publish a summary of the content used for training. The most powerful models, those with systemic risk, carry extra duties: evaluations, risk management, serious-incident reporting and cybersecurity.
To help companies, the Commission published a voluntary Code of Practice on 10 July. Signatories include OpenAI, Anthropic, Google, Microsoft, Amazon, IBM, Mistral and Aleph Alpha; xAI signed only the safety part; Meta refused. Fines do not come immediately though: the Commission can sanction from 2 August 2026, and models already on the market before 2 August 2025 have until 2 August 2027.
Companies
European Commission, EU AI Office
Tools
EU AI Act GPAI Code of Practice
Tags
Sources