The European Commission has published the third and final draft of the General-Purpose AI Code of Practice, a key regulatory framework intended to ensure compliance with the EU AI Act. The draft, refined through extensive stakeholder engagement, introduces structured commitments focusing on transparency, copyright, safety, and security, particularly for AI models with systemic risks.
The final draft aims to provide AI model providers with a practical blueprint for meeting the obligations outlined in the AI Act, which comes into effect in August 2025. Stakeholders have until March 30, 2025, to submit feedback before the Code is finalized in May.
The third draft of the Code is structured into three core sections, each with a distinct regulatory focus:
- Transparency commitments: AI providers must document and disclose essential model information, including capabilities, limitations, and resource consumption. A newly introduced Model Documentation Form streamlines the process, ensuring AI Office and regulatory bodies can access necessary data when required.
- Copyright and data use: AI providers must clarify how copyrighted content is used for training, offering simplified legal protections. The draft strengthens measures ensuring compliance with EU copyright laws while allowing AI development to progress within legal constraints.
- Safety and security for high-risk AI models: Providers of AI models classified as General-Purpose AI with Systemic Risk must implement systemic risk assessments, mitigation strategies, and cybersecurity protocols. The draft highlights best practices, such as automated red-teaming and human oversight, to ensure AI systems operate safely.
A significant update in the new draft acknowledges the rapid evolution of AI technology. The framework ensures flexibility, allowing future adaptation to emerging risks and innovations. To facilitate this, the EU AI Office will provide additional guidance later in 2025 on defining general-purpose AI models, provider responsibilities, and the application of exemptions for open-source models.
The Code has been developed through an inclusive process, involving 1,000 stakeholders from the AI industry, academia, and regulatory bodies. The AI Office will now host working group discussions and workshops to gather final feedback before adopting the Code.
Following the consultation phase, the finalized General-Purpose AI Code of Practice will serve as a voluntary compliance tool to help AI model providers align with the EU AI Act, reinforcing Europe’s commitment to responsible AI governance.
Need Help?
If you’re concerned or have questions about how to navigate Ireland’s, or any country’s, AI regulatory landscape, don’t hesitate to reach out to BABL AI. Their Audit Experts can offer valuable insight and ensure you’re informed and compliant.
