On 10-7-2025, the European Commission announced the ‘General-Purpose AI Code of Practice’, marking a pivotal step in Europe’s approach to Artificial Intelligence (AI) governance. The Code offers a voluntary framework to help providers comply with the Europe’s (‘EU’) AI regulations and reflects the input and priorities of key AI stakeholders. The Code will come into effect on 2-8-2025.
Background:
- The General-Purpose AI (‘GPAI’) Code of Practice emerged as a voluntary compliance tool under the EU’s Artificial Intelligence (EU AI) Act (Regulation EU 2024/1689)1, which was formally signed in June 2024 and entered into force on 1-8-2024.
- Designed to guide providers of general-purpose AI models, the Code is drafted by 13 independent experts with input from over 1,000 stakeholders, including developers, small and medium-sized enterprises, researchers, and civil society.
- Working groups and providers helped refine practical instruments such as model documentation templates and risk assessment protocols.
-
The final version was reviewed by the Commission’s AI Committee in July 2025, ensuring its compliance with legal standards and uses the latest best practices.
Key Takeaways of AI Code of Practice:
- The Code of Practice outlines measures aligned with the EU AI Act (2024) to promote safety and transparency across the European AI landscape,
- The Code is structured into three chapters:
-
Transparency
-
Copyright
-
Safety & Security
-
- The Transparency chapter introduces a user-friendly Model Documentation Form to help providers record key information clearly and efficiently, in a single place.
- The Copyright chapter offers practical strategies for aligning with EU copyright law.
- The Safety & Security chapter targets high-impact models with guidance on managing systemic risks such as misuse or loss of control.
- These chapters are aligned with the obligations of Article 53 and Article 55 of EU AI Act.
- Member State and Commission endorsement is the next key step, after which Commission Guidelines will follow to clarify the scope of GPAI obligations, ensuring providers know exactly where they stand under the AI Act.
-
Signatories to the Code will benefit from reduced administrative burdens and greater legal certainty compared to those using alternative compliance methods.
GPAI Code of Practice is expected to ease the implementation of the EU AI Act by offering providers a ready-to-use, stakeholder-informed compliance framework. It paves way for safer, more transparent deployment of general-purpose AI models across Europe, with full enforcement starting in August 2026 for new GPAI models and August 2027 for existing GPAI models.
