A blueprint for democratic governance of frontier AI

This document proposes a strategy for the U.S. government to create lasting institutions that oversee frontier AI (the most advanced AI systems being developed). The plan has three main parts: build a national framework based on state laws already in place, strengthen CAISI (the federal organization responsible for frontier AI safety) as the main federal institution, and develop a broader government-wide plan to address national security and public safety risks from advanced AI.