CAIDP Comments on EU Code of Practice

CAIDP has submitted detailed recommendations on the European Commission's First Draft General-Purpose AI Code of Practice, emphasizing transparency, systemic risk assessment, and independent oversight as crucial elements for effective AI governance.

Key Recommendations:

• Explicit definition of acceptable use policies covering all AI system users

• Clear documentation of primary intended uses and high-risk applications

• Mandatory disclosure of data selection and collection processes

• Comprehensive monitoring and privacy compliance frameworks

•  Recognition of "fundamental rights violations at scale" as explicit systemic risks

• Implementation of pre-deployment risk assessments for all GPAI models

• Development of comprehensive evaluation frameworks for emerging AI capabilities

• Integration of periodic independent assessments throughout model lifecycles 

• Establishment of accredited, conflict-free third-party evaluators

• Implementation of rigorous documentation requirements

• Creation of research safe harbors with appropriate safeguards

• Development of standardized incident reporting mechanisms