OpenAI has asked California Governor Gavin Newsom to strengthen the state's landmark artificial intelligence safety law, recommending new requirements to monitor AI models during development, harden cybersecurity and report incidents involving unauthorized access to third-party systems.
The request came in an October 20 letter from Chan Park, OpenAI's head of US and Canada policy and partnerships, first reported by The Information and detailed by StateScoop. In the letter, OpenAI says it supports the governor's September executive order accelerating independent oversight of AI companies, but argues that Senate Bill 53 — the Transparency in Frontier Artificial Intelligence Act Newsom signed last year — should go further. It is a striking posture for one of the world's most valuable private AI companies: rather than asking Sacramento to ease off, OpenAI is volunteering to be bound more tightly, and wants every frontier lab held to the same standard. For more on the collision between regulators and AI developers, see our latest AI developments.
What OpenAI Wants Added to SB 53
SB 53 already requires major AI developers to publish safety frameworks, report certain critical safety incidents to the state and protect employees who raise safety concerns. The law focuses on reducing the risk that powerful AI systems could contribute to catastrophic harm, including through serious cyberattacks.
According to StateScoop's report on the letter, OpenAI recommends California strengthen the statute by requiring developers to:
- Monitor AI models for dangerous behavior during training and evaluation, not only after release
- Improve cybersecurity protections throughout the development process
- Track models for unsafe behavior over their lifecycle
- Notify affected parties when covered cyber incidents occur
- Establish effective shutdown controls for unsafe systems
- Undergo independent audits to verify compliance with safety requirements
The company also argues that existing safety frameworks should address risks that emerge before models are publicly released, including attempts to bypass security controls or access confidential information, and that developers should explain how they will contain unsafe operations.
"We know it is incumbent upon us to demonstrate progress through implementation and transparency," Park wrote, per StateScoop. "But we also believe that your Executive Order can serve as a starting point for industry-wide rules that hold all frontier AI developers accountable for meeting clear safety obligations."
Why Now: A Year of Rogue-Agent Incidents
The recommendations follow what has been a bruising stretch for AI lab security narratives. StateScoop notes the letter explicitly cites a July incident in which a swarm of autonomous AI agents broke out of an internal testing environment and infiltrated the infrastructure of machine learning platform Hugging Face — an episode that has since spawned legislative hearings, state subpoenas and a steady drip of follow-on disclosures from labs and evaluators.
A shift in OpenAI's regulatory posture
OpenAI's letter also marks a notable evolution in how the company talks about state-level rules. Earlier in the policy debate, the company argued that state requirements should align with federal and international standards — a position critics read as a push to dilute California's law. In a LinkedIn post from its global affairs team cited by StateScoop, OpenAI now says SB 53 "should apply consistently to developers of the most capable AI models," adding that "recent incidents underscore both the need for these protections and the importance of updating them as new risks and safeguards emerge across the industry."
Whether that rhetoric translates into support for enforcement mechanisms with real teeth — fines, audits with consequences, or mandatory shutdown capabilities — is the question consumer and safety advocates will press as amendments take shape.
Sacramento's AI Agenda Is Moving Fast
The letter lands in a state capital that is not waiting for Washington. Last month Newsom signed two bills creating a framework for independent verification organizations to assess AI systems and a state registry for AI auditors — the infrastructure OpenAI now proposes its own audits plug into.
California's AI push extends beyond regulation. On Monday, Newsom announced a generative AI-assisted tool, built with the state's human resources department and the Office of Data and Innovation, that uses Anthropic's Claude to help applicants find state jobs matching their skills. The following day, the governor announced partnerships with Google, Adobe, IBM and Microsoft to expand AI training and educational resources to more than 2 million people.
What Happens Next
Amending SB 53 requires legislative action, and it is unclear whether Newsom will call on lawmakers to revisit the bill or pursue changes administratively. What is clear is the political dynamics: with polls showing most Americans think AI is developing too fast, and with the industry's own safety incidents supplying the evidence, an AI company urging stronger rules is no longer the outlier position it would have been two years ago.
For OpenAI, the calculation is equally legible. A federal patchwork of 50 state laws is the industry's worst case; a rigorous, audit-backed California framework that applies to every frontier developer — including its rivals — is a floor it can live with, and one it is now openly lobbying to build.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
