The European Commission has advanced its efforts to regulate artificial intelligence with the release of the second draft of the General-Purpose AI (GPAI) Code of Practice. This draft reflects continued progress towards aligning AI development with the EU Artificial Intelligence Act (AI Act), which will take effect in August 2025.
The new draft refines the framework guiding AI providers in meeting the AI Act’s requirements. It sharpens obligations around transparency and copyright, with specific exemptions for certain open-source models to balance innovation with risk management. Providers of advanced GPAI models identified as posing systemic risks will be subject to more rigorous standards, including mandatory risk assessments, model evaluations, and incident reporting. Cybersecurity provisions are also highlighted, reinforcing the EU’s focus on safeguarding AI systems.
The drafting process draws from broad engagement with stakeholders, including nearly 1,000 participants from AI developers, industry bodies, civil society, and member states. The second draft integrates feedback from working groups and provider workshops held in November 2024, alongside 354 written submissions. Insights from interinstitutional meetings with the European AI Board and European Parliament have also shaped the evolving document.
Key elements of the draft focus on transparency and copyright obligations for GPAI providers, requiring disclosures about training data and adherence to copyright standards. Advanced models identified as having potential systemic impact are subject to risk management protocols and cybersecurity measures. The Code now adopts a modular structure that sets objectives and performance indicators, providing clarity while allowing flexibility for technological advancements.
Verbal discussions and additional feedback sessions will continue into January 2025, with the third draft expected by mid-February. The final version is anticipated by April 2025, positioning the EU to implement the Code ahead of the AI Act’s enforcement.
As the EU progresses toward finalizing the GPAI Code of Practice, this initiative shows Europe’s leadership in AI governance. By fostering collaboration and emphasizing accountability, the Commission seeks to establish a robust framework that promotes responsible AI innovation and deployment across the region.
About the author:
