The AI policy window is open. We need to act.

OpenAI argues that a critical policy window for governing advanced AI is currently open and must be utilized urgently. The company calls for mandatory, capability-based national safety regulation in the U.S., advocating for legislation before Congress adjourns. Until federal action is taken, OpenAI supports specific state-level bills in California that address independent safety assessments, auditor standards, youth protections, and biological threat safeguards. The article emphasizes that industry-led standards, particularly for monitoring frontier models for misalignment and serious incidents, are necessary complements to government regulation. It stresses the need for international compatibility in standards and warns against inaction, stating that safety and trust are prerequisites for widespread adoption and American technological leadership, not impediments to progress. The core message is that policy must evolve in tandem with rapid AI advancements, including the potential for recursive self-improvement, and that stakeholders should act decisively now.

(Source：OpenAI)