Glossary · EU AI Act: general-purpose AI and transparency
GPAI Code of Practice
Also known as: General-Purpose AI Code of Practice, code of practice for general-purpose AI
The General-Purpose AI Code of Practice is a voluntary code, published by the European Commission on July 10, 2025 and developed by independent experts with stakeholders, that helps providers of general-purpose AI models show compliance with their obligations under the EU AI Act in three chapters: transparency, copyright, and safety and security.
- AI regulation
- EU
In one sentence
The GPAI Code of Practice (July 2025) is a voluntary code helping general-purpose AI model providers comply with the AI Act.
Example
A model provider that signs the code uses its model documentation form to give downstream providers details on training data, capabilities and limitations.
How it applies
- Chapters: Transparency (including a model documentation form) and Copyright apply to all GPAI providers; Safety and Security applies only to providers of models with systemic risk.
- Status: The code is voluntary. Signatories can demonstrate compliance by adhering to it, which the Commission says brings more legal certainty and less administrative burden; providers who do not sign must show compliance by other means.
- Technical documentation: The model documentation form is a practical template for the information GPAI providers owe the AI Office and downstream providers.
Compared with the USA, Canada and China
Canada's voluntary code of 2023 is the closest counterpart, but it is not linked to binding law. In the USA, the voluntary commitments that leading AI companies made to the White House in 2023 have no binding force; California's SB 53 turned similar safety and transparency practices into law for frontier developers. China uses technical standards from its national cybersecurity standardization committee (TC260) in a similar role.