This document is the first draft of the General-Purpose AI Code of Practice under the EU AI Act. It was developed through a multi-stakeholder consultation process, incorporating nearly 430 submissions from industry, academia, and civil society. The draft outlines transparency, risk identification, technical and governance risk mitigation measures for providers of general-purpose AI models, especially those with systemic risks. The document establishes principles for alignment with EU values, proportionality to risks, and fostering an AI safety ecosystem. It highlights open questions for stakeholder feedback and emphasizes a commitment to iterative refinement and collaboration. The final version is expected by May 1, 2025, to guide compliance and governance in the evolving landscape of AI development and deployment.
Authors
- Published in
- Belgium
