General-purpose AI (GPAI) model

A general-purpose AI (GPAI) model is, in the words of Article 3(63) of Regulation (EU) 2024/1689, the EU AI Act, an AI model that "displays significant generality and is capable of competently performing a wide range of distinct tasks" and can be integrated into a variety of downstream systems or applications. The large language models behind chat assistants and copilots are the typical example. The Act regulates the model itself, separately from the AI systems built on it.

Chapter V of the Act has applied to GPAI model providers since 2 August 2025. Under Article 53, a provider must keep technical documentation, give downstream providers the information they need to use the model and meet their own obligations, put in place a policy to comply with EU copyright law, and publish a sufficiently detailed summary of the content used for training; some open-source models are exempt from the first two duties. Models with systemic risk, presumed when the compute used for training exceeds 10^25 floating-point operations (Article 51), carry further duties under Article 55: model evaluations including adversarial testing, assessment and mitigation of systemic risks, serious incident reporting and cybersecurity protection. Models placed on the market before 2 August 2025 have until 2 August 2027 to comply (Article 111(3)).

The Commission published a voluntary General-Purpose AI Code of Practice on 10 July 2025, with chapters on transparency, copyright, and safety and security, as one way for providers to show compliance. Since 2 August 2026 the Commission's AI Office, together with national authorities, has been enforcing the Act.

For most compliance teams, the GPAI rules are a vendor diligence question: using a tool built on a GPAI model does not make you its provider, but modifying a model and supplying it to others under your own name can. The EU AI Act checklist for deployers covers where that line sits.

This entry is general information, not legal advice.