OpenAI Calls for Mandatory Federal AI Safety Rules and Backs State Legislation

In a policy post published on September 9, 2026, OpenAI Chief Global Affairs Officer Chris Lehane called for mandatory, capability-based national safety regulations and endorsed several state legislative efforts.

OpenAI outlined four primary policy priorities:

• Working with Congress on mandatory national AI safety requirements
• Supporting state-level safety legislation
• Advancing industry-led standards across frontier laboratories
• Advocating for compatible global standards on risk management and development thresholds

The company stated that federal requirements should focus specifically on well-resourced laboratories building the most capable models, rather than startups or non-frontier developers. OpenAI's proposed framework includes independent assessments, incident reporting, cybersecurity standards, and shared metrics to monitor progress toward recursive self-improvement. While fully autonomous recursive self-improvement is not occurring today, OpenAI noted that AI agents can already perform research tasks that would take skilled researchers several days.

Until Congress enacts federal rules, OpenAI is supporting state bills to create baseline protections. The company endorsed four California bills: SB 813 for independent safety assessment infrastructure, AB 1405 for AI-auditor standards, SB 1119 for youth protections, and AB 1864 for biological threat safeguards. It noted previous backing for California SB 53, New York's RAISE Act, Illinois SB 315, and independent audit provisions under consideration in Massachusetts.

The post also detailed internal safeguards for its model Astra, including full trajectory monitoring of chains of thought and an alignment-evaluation gate prior to broad internal deployment. OpenAI stated it will slow or stop development or deployment under its preparedness framework whenever systems cannot be sufficiently safeguarded.