Summary

Policy brief arguing that evaluation governance is the missing regulatory layer. Current AI safety regulation mandates evaluation without governing the evaluation methodology itself, creating systematic blind spots.