AI Frontier Institute

Artificial intelligence models: large developers.

View primary source →

Senate Bill 53 (Chapter 138), signed September 29, 2025, enacts the Transparency in Frontier Artificial Intelligence Act (TFAIA) and related provisions across three California codes. Under the Business and Professions Code (Chapter 25.1, Sections 22757.10–22757.16), 'large frontier developers'—defined as companies with over $500 million in annual gross revenue that have trained a 'frontier model' (a foundation model trained using more than 10^26 integer or floating-point operations)—must: (1) write, implement, and publicly post a 'frontier AI framework' describing how they assess and mitigate catastrophic risks, reviewed at least annually; (2) publish transparency reports before or concurrently with deploying new or substantially modified frontier models; (3) transmit quarterly summaries of internal catastrophic-risk assessments to the California Office of Emergency Services; and (4) refrain from making materially false or misleading statements about catastrophic risk or their frontier AI framework. 'Catastrophic risk' is defined as a foreseeable, material risk of death or serious injury to more than 50 people or more than $1 billion in property damage from a single incident involving CBRN weapon assistance, autonomous harmful conduct without meaningful human oversight, or loss of developer control. Frontier developers (not only large ones) must report 'critical safety incidents' to the Office of Emergency Services within 15 days of discovery, or within 24 hours if the incident poses imminent risk of death or serious physical injury. The Office of Emergency Services must establish reporting mechanisms, protect submitted reports from public disclosure (exempting them from the California Public Records Act), and produce anonymized annual public reports beginning January 1, 2027. Civil penalties up to $1 million per violation, enforceable only by the Attorney General, apply to noncompliance (Section 22757.15). Under the Government Code (Section 11546.8), a 14-member consortium is established within the Government Operations Agency to develop a framework for 'CalCompute,' a public cloud computing cluster intended to advance safe, ethical, equitable, and sustainable AI; the consortium must submit a report to the Legislature by January 1, 2027, after which it dissolves. This provision is operative only upon an appropriation. Under the Labor Code (Chapter 5.1, Sections 1107–1107.2), frontier developers are prohibited from retaliating against or silencing 'covered employees' (those responsible for assessing, managing, or addressing critical safety incident risk) who report catastrophic risks or TFAIA violations to the Attorney General, federal authorities, or supervisors. Large frontier developers must provide an anonymous internal reporting process with monthly status updates to whistleblowers. Successful plaintiffs may recover attorney's fees, and burden of proof shifts to the employer once a protected activity is shown to be a contributing factor to retaliation. The act preempts local government regulation of frontier developers regarding catastrophic risk management adopted on or after January 1, 2025, and defers to conflicting federal law or contracts.

← Back to the tracker