CA SB53: Artificial intelligence models: large developers.

In one paragraph

California's SB 53, the Transparency in Frontier Artificial Intelligence Act, covers developers of frontier models (foundation models trained with more than 10^26 operations). Large frontier developers (over $500 million in prior-year revenue with affiliates) must publish and follow a frontier AI framework for catastrophic risks. All frontier developers must publish transparency reports at deployment, report critical safety incidents to the Office of Emergency Services within 15 days, and not retaliate against whistleblowers. Large developers face penalties up to $1 million per violation. Signed September 29, 2025 (Chapter 138); effective January 1, 2026.

Plain-English summary, written with AI assistance. The official text controls.

What it means for business

Frontier model developers (over 10^26 training operations) must publish transparency reports, report critical safety incidents within 15 days and protect whistleblowers; those with over $500M in revenue must also follow a published frontier AI framework and face penalties up to $1M per violation.

Key provisions

1. 1Definitions (Bus. & Prof. Code § 22757.11): a frontier model is a foundation model trained using more than 10^26 integer or floating-point operations, counting the original training run plus later fine-tuning, reinforcement learning or other material modifications; a large frontier developer is a frontier developer whose annual gross revenues, with its affiliates, exceeded $500,000,000 in the preceding calendar year
2. 2Catastrophic risk: a foreseeable and material risk that a frontier model will materially contribute to the death of or serious injury to more than 50 people, or more than $1 billion in property damage, from a single incident in which the model gives expert-level help creating or releasing a chemical, biological, radiological or nuclear weapon, carries out a cyberattack or murder-, assault-, extortion- or theft-like conduct without meaningful human oversight, or evades its developer's or user's control
3. 3Large frontier developers must write, implement, comply with and publish a frontier AI framework (standards, risk thresholds, mitigations, third-party assessments, security of unreleased model weights, incident response, governance, internal-use risks), review it at least yearly, publish material changes with a justification within 30 days, and send OES summaries of internal-use catastrophic-risk assessments every three months or on another reasonable schedule (§ 22757.12)
4. 4Before or when deploying a new or substantially modified frontier model, every frontier developer must publish a transparency report (website, contact mechanism, release date, languages, output modalities, intended uses, use restrictions); large frontier developers add summaries of their catastrophic-risk assessments, results and third-party evaluators' role. Materially false or misleading statements about catastrophic risk or framework compliance are barred unless made in good faith and reasonable; redactions are allowed for trade secrets, cybersecurity, public safety, national security or legal compliance, with unredacted copies kept five years
5. 5Frontier developers must report a critical safety incident (e.g., unauthorized access to model weights or loss of control causing death or bodily injury, or a model using deception to subvert its developer's controls in a way showing materially increased catastrophic risk) to OES within 15 days of discovery, and within 24 hours to an appropriate authority such as law enforcement if it poses an imminent risk of death or serious physical injury; these reports are exempt from the Public Records Act (§ 22757.13)
6. 6Whistleblowers (Labor Code § 1107 and following): frontier developers may not prevent or retaliate against covered employees (those responsible for assessing, managing or addressing critical safety incident risk) who, with reasonable cause, tell the Attorney General, a federal authority or a supervisor of a specific and substantial danger from catastrophic risk or a violation, and must notify them of their rights; large frontier developers must offer an anonymous internal channel with monthly status updates; successful plaintiffs may recover attorney's fees
7. 7Large frontier developers that fail to publish or transmit required documents, make prohibited statements, fail to report incidents or do not follow their own framework face civil penalties up to $1,000,000 per violation, depending on severity, in actions brought only by the Attorney General (§ 22757.15); local rules on frontier developers' management of catastrophic risk adopted on or after January 1, 2025 are preempted; a CalCompute consortium is to report a framework for a public cloud computing cluster by January 1, 2027, operative only upon an appropriation (Gov. Code § 11546.8)
8. 8Signed by the Governor September 29, 2025 (Chapter 138, Statutes of 2025); effective January 1, 2026. Annual OES and Attorney General reports and Department of Technology reviews of the definitions begin January 1, 2027

Latest legislative action

Sep 29, 2025. Chaptered by Secretary of State. Chapter 138, Statutes of 2025.

Sponsors (showing 5 of 78)

Roll call votes (showing 3 of 20)