CAIDP Submits Key Recommendations on EU AI Code of Practice

CAIDP has submitted comprehensive recommendations to strengthen the European Union's second draft General-Purpose AI Code of Practice, emphasizing the need for mandatory commitments and transparent oversight mechanisms. The Center's analysis highlights concerns about the current draft's reliance on voluntary measures, noting that phrases like "are encouraged" appear 30 times.

Key recommendations include replacing permissive language with firm requirements, making providers' risk assessments publicly available, and establishing independent evaluation protocols. CAIDP also advocates for explicit recognition of "violations of fundamental rights at scale" as a systemic risk and supports the development of standardized templates for documentation.

The recommendations emphasize the importance of public transparency in AI development, particularly regarding environmental impact assessments and risk evaluations. CAIDP continues its active participation in Working Groups 1, 2, and 4 of the drafting process.

As February 2, 2025 approaches, the first major milestone of the EU AI Act will take effect, implementing key prohibitions on harmful AI practices, establishing foundational definitions, and introducing measures to promote AI literacy.