Use this image only if you have the right to publish it or it is licensed for your use.

OpenAI has called for California to strengthen parts of SB 53, the state’s frontier-AI safety law, according to a post from the company’s global affairs team reported by TechCrunch. The company said the law should be amended to add safeguards during the development of powerful AI models, including monitoring while models are being trained or evaluated and stronger cybersecurity protections throughout the development lifecycle.

The request matters because SB 53 is already a significant state-level framework. California signed the measure into law in September 2025 as the Transparency in Frontier Artificial Intelligence Act. The law requires large frontier-model developers to publish safety frameworks, creates a mechanism for reporting potential critical safety incidents, and includes whistleblower protections and civil penalties for noncompliance.

What OpenAI is asking California to consider

TechCrunch reported that OpenAI wants the law expanded rather than replaced. The company’s proposals focus on two practical questions: whether serious incidents should be monitored earlier, during model training or evaluation, and whether cybersecurity requirements should be strengthened across the full development process.

Those proposals are different from simply asking companies to publish a policy after a model is released. They would place more attention on how developers identify and respond to risks before a system reaches the public. Whether California adopts any amendment will depend on the legislature and the governor; OpenAI’s statement is a policy position, not a change to the law.

Why SB 53 is already important

California describes SB 53 as a law designed to balance innovation with public-safety guardrails for frontier AI. Its existing requirements include public transparency frameworks, reporting channels for critical safety incidents and protection for workers who disclose significant health or safety risks.

That makes the debate more specific than a broad argument about whether AI should be regulated. The real question is how detailed the requirements should be, when they should apply and how much evidence a developer should provide when assessing potential risks.

A notable change in OpenAI’s position

OpenAI’s intervention drew attention because the company had previously opposed SB 53, TechCrunch reported. The company now says states can move in compatible directions around core safeguards while federal legislation remains limited. That does not end the policy debate, but it does show how quickly the discussion has shifted from whether a state framework should exist to how strong its requirements should be.

For developers, users and policymakers, the next point to watch is whether California proposes a formal amendment. Until then, SB 53 remains the operative framework, and OpenAI’s recommendations should be understood as a public request for additional safeguards rather than rules that have already taken effect.

Sources

• Governor of California — SB 53 signed into law
• TechCrunch — OpenAI says California should strengthen its AI safety bill

Last checked: August 25, 2026. This article summarizes public policy reporting and is not legal advice. For a correction or source question, contact contact.globalledger@gmail.com.