Legal and Policy Approaches to Mitigate Catastrophic Harms from AI

A RAND study surveyed 24 experts on policies to prevent catastrophic AI risks and found alarmingly few viable solutions. The experts' preference for voluntary standards and disclosure incentives over mandatory regulation suggests current policy frameworks may be inadequate to address existential-level AI threats. The reliance on state and industry self-regulation for such high-stakes risks is particularly concerning.