OpenAI, Anthropic urge Newsom to tighten California AI safety rules
The two AI companies sent separate letters to Governor Gavin Newsom, differing on whether regulators should be able to force a full model shutdown.
OpenAI and Anthropic each sent letters to California Governor Gavin Newsom calling for stronger AI safety regulation, according to Reuters, but the two companies took different positions on mandatory kill switches.
Anthropic's letter, sent on the 15th, said the government should be able to ask a court to block the release of an AI model found to pose serious risk. The company opposed mandatory kill switches that would shut down an entire model or system at once, arguing that targeted interventions, such as disabling a problematic AI agent, are more effective.
OpenAI's letter, sent on the 20th, proposed requiring AI companies to have emergency shutdown controls, undergo annual independent external audits and strengthen cybersecurity standards. The company said safety frameworks should explicitly address advanced models' ability to automate AI research and development.
OpenAI proposed strengthening California's SB53 AI transparency law by adding emergency shutdown controls and independent audits to its existing obligations. The law, enacted in 2025, requires advanced AI developers to publish safety frameworks and transparency reports and to report serious safety incidents.
In September 2026, Newsom formed an expert panel to prepare improvements to AI safety regulation. When SB53 was debated in the California legislature, OpenAI opposed what it described as excessive state-level regulation.
Quick answers
What did OpenAI and Anthropic ask Gavin Newsom to do?
Both companies sent letters urging stronger AI safety regulation in California, though they differed on mandatory kill switches.
What is SB53?
California's 2025 AI transparency law, which requires advanced AI developers to publish safety frameworks and transparency reports and to report serious safety incidents.