European Commission Proposes Risk-Based AI Regulation, Banning ‘Unacceptable’ Uses
April 22, 2021
The European Commission released a long-awaited proposal for how to regulate AI throughout the European Union. The proposed regulation includes a ban on “unacceptable” uses of AI such as general social scoring and “real time remote biometric identification” for law enforcement. The proposal also imposes testing and transparency obligations for "high-risk" uses of AI, including a publicly accessible EU database on stand-alone “high-risk” systems. The proposal requires notice to individuals when they interact with certain types of AI and “conformity” assessments for "high-risk" systems. The prohibitions on unacceptable AI are very limited and many of the strongest provisions are subject to vast exceptions. However, a penalty of up to 4% of annual revenue on companies that violate the regulation is included. EPIC has called for prohibitions on secret scoring, mass surveillance, and facial recognition. EPIC urges legislators to implement the OECD Principles on AI and adopt the Universal Guidelines of AI.