As the implementation of the EU AI Act begins, a crucial mechanism for enforcing regulations on General Purpose AI (GPAI) models is taking shape: the Code of Practice. This interim measure bridges the gap between the Act's enforcement and the development of harmonized European standards.
The Code of Practice for GPAI model providers is a critical component of the EU AI Act's implementation. It serves as a temporary compliance mechanism until formal European standards are developed. Key points include:
1. Purpose: To operationalize obligations for GPAI model providers before official standards are in place.
2. Timeline: Must be drafted within nine months of the AI Act's entry into force (by May 1, 2025).
3. Process: Involves a multi-stakeholder Plenary divided into four working groups, focusing on transparency, risk assessment, risk mitigation, and governance.
4. Participation: Open to various stakeholders, including GPAI model providers, downstream providers, academia, and civil society organizations.
5. Leadership: Chaired by independent experts who will guide the drafting process.
6. Content: Expected to cover technical documentation, information sharing, training data summaries, copyright compliance, and additional measures for high-risk models.
7. Compliance: While not legally binding, adherence to the Code will presume conformity with AI Act obligations.
The Code of Practice represents a crucial step in implementing the AI Act, balancing the need for rapid action with the complexities of regulating emerging technologies.
Read the full Code of Practice Explainer here.
For questions, please reach out to our EU AI policy lead Jimmy Farrell at jimmy.farrell@pourdemain.eu.
Comments