California frontier AI transparency law (SB 53, 2025)
A California law of September 2025 that requires large developers of frontier AI models to publish their safety framework and report critical safety incidents.
The most powerful models are developed by a handful of companies, many headquartered in California, which gives that state's law a de facto global reach. After the veto of a first, more stringent bill in 2024, the law takes a transparency approach: publishing how the company assesses and mitigates catastrophic risks, reporting critical safety incidents to the state authority, protecting internal whistleblowers. It sets no threshold of acceptable risk, but it creates a written record against which the developer's conduct can later be judged. That record is what the liability insurer will be able to read, and what a plaintiff will be able to invoke.
Governor Gavin Newsom signed SB 53 on September 29, 2025, a year after vetoing, on September 29, 2024, bill SB 1047, which notably provided for testing obligations and an emergency shutdown mechanism.
Follow the research
One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.
No spam, one-click unsubscribe, no data ever sold.