BETA RELEASE

Summary

A proposal for regulating frontier AI models through standard-setting, registration, and compliance mechanisms to manage risks to public safety.

Key quotes

Frontier AI models pose a distinct regulatory challenge: dangerous capabilities can arise unexpectedly; it is difficult to robustly prevent a deployed model from being misused; and, it is difficult to stop a model's capabilities from proliferating broadly.

The paper discusses the need for government intervention and specific safety standards for highly capable foundation models. Proposed standards include pre-deployment risk assessments and external scrutiny of model behavior.