Glossary · United States
California SB 53 (Transparency in Frontier AI Act)
Also known as: Transparency in Frontier Artificial Intelligence Act, TFAIA, SB 53
California SB 53, the Transparency in Frontier Artificial Intelligence Act, is a California law, signed on September 29, 2025 and in effect since January 1, 2026, that requires developers of frontier AI models to publish safety frameworks and transparency reports, report critical safety incidents to the state and protect whistleblowers.
- AI regulation
- USA
- Technical documentation
In one sentence
California SB 53 (in effect since January 2026) requires frontier AI developers to publish safety frameworks and transparency reports and report incidents.
Example
Before releasing a new frontier model, a large developer publishes a transparency report summarizing its catastrophic-risk assessment and the role of third-party evaluators.
How it applies
- Scope: Frontier developers — those who trained a model with more than 10²⁶ integer or floating-point operations. Large frontier developers, with annual revenues above USD 500 million, have additional duties.
- Duties: Large frontier developers publish a frontier AI framework describing how they assess and mitigate catastrophic risks and apply standards. Before deploying a new or substantially modified frontier model, developers publish a transparency report. Critical safety incidents must be reported to the California Office of Emergency Services within 15 days, or within 24 hours to an appropriate authority if there is an imminent risk of death or serious injury.
- Enforcement: Civil penalties of up to USD 1 million per violation, enforced by the Attorney General. The law also protects whistleblowers and creates a consortium to design a public computing cluster, CalCompute.
- Technical documentation: The framework and transparency reports are public documents. They must be consistent with what the company actually does — false or misleading statements are themselves violations.
California SB 53 vs. the EU AI Act
SB 53 covers the same ground as the AI Act's rules for general-purpose AI models with systemic risk but for fewer companies (10²⁶ instead of 10²⁵ operations) and with publication instead of documentation for an authority. Many frontier developers align their SB 53 framework with the Safety and Security chapter of the EU GPAI Code of Practice.