Transparency in Frontier Artificial Intelligence Act (TFAIA)

Summary

California SB 53 — the Transparency in Frontier Artificial Intelligence Act (TFAIA), also searched as "CA SB 53", "California frontier AI law", or "frontier AI safety act" — was signed by Governor Newsom on September 29, 2025 after the prior frontier-AI bill (SB 1047) was vetoed in 2024. It is codified across Business and Professions Code Chapter 25.1 (Sections 22757.10 through 22757.16), Government Code Section 11546.8, and Labor Code Chapter 5.1 (Sections 1107 through 1107.2), with the core frontier-developer provisions effective January 1, 2026. The law imposes published safety-framework, model-transparency, incident-reporting, and whistleblower-protection duties on developers of the largest AI models.

SB 53 is the frontier-model tier of California's generative-AI transparency regime, which also includes California SB 942, the AI Transparency Act, governing disclosure of AI-generated content, and California AB 2013, the Generative AI Training Data Transparency Act requiring developers to publish training-data documentation.

Who must comply with SB 53: frontier developers vs. large frontier developers

SB 53 applies only to developers of the largest models, sorted into two tiers by a compute threshold and a revenue threshold. The distinction controls which obligations attach, so scoping starts here:

• Frontier model — a foundation model trained using a quantity of computing power greater than 10^26 (one hundred septillion) integer or floating-point operations, including the compute used in any later fine-tuning.
• Frontier developer — a person that has trained, or initiated the training of, a frontier model. Frontier developers carry the transparency-report and critical-safety-incident-reporting duties.
• Large frontier developer — a frontier developer that, together with its affiliates, had annual gross revenue exceeding $500,000,000 in the preceding calendar year. Only large frontier developers must publish and maintain a frontier AI framework and the catastrophic-risk assessment summaries.

Frontier AI safety framework requirement

A large frontier developer must write, implement, comply with, and publish on its website a frontier AI framework describing how the organization assesses and mitigates catastrophic risk from its frontier models. The framework must be reviewed and updated at least once per year, and material modifications must be republished.

Transparency report requirement

Before or concurrently with deploying a new frontier model, a frontier developer must publish a transparency report covering the model and its intended uses. A large frontier developer must additionally include summaries of its catastrophic-risk assessments and the steps taken to mitigate those risks.

Critical safety incident reporting to Cal OES

A frontier developer must report any critical safety incident involving one or more of its frontier models to the California Office of Emergency Services within 15 days of discovering the incident. If a critical safety incident poses an imminent risk of death or serious physical injury, the developer must disclose it within 24 hours to an appropriate authority — including a law enforcement or public safety agency with jurisdiction — based on the nature of the incident.

Whistleblower protections for covered employees

Frontier developers must preserve the whistleblower rights of covered employees who raise catastrophic-risk concerns, and a large frontier developer must maintain a reasonable anonymous internal process through which covered employees can disclose those concerns. Retaliation against a protected disclosure is prohibited under Labor Code Sections 1107 through 1107.2.

CalCompute public computing cluster

The Government Operations Agency must develop a framework for CalCompute, a proposed public cloud-computing cluster intended to expand access to compute for safety-focused and academic AI research. The CalCompute provisions become operative only upon an appropriation in a budget act or other statute.

SB 53 penalties and enforcement

The California Attorney General enforces the TFAIA. Under Business and Professions Code Section 22757.15(a), a large frontier developer that fails to publish or transmit a compliant document, makes a statement in violation of Section 22757.12(e), fails to report an incident as required by Section 22757.13, or fails to comply with its own frontier AI framework is subject to a civil penalty — set by the severity of the violation — not exceeding one million dollars ($1,000,000) per violation.

How SB 53 differs from SB 1047

SB 53 is narrower than the vetoed SB 1047 (2024). It centers on transparency — published safety frameworks, transparency reports, and incident reporting — rather than the pre-deployment liability, third-party audits, and full-shutdown ('kill switch') capability that SB 1047 would have required. SB 53 imposes no general civil liability for catastrophic harm; its penalties attach to disclosure and framework-compliance failures.

_Primary sources: the chaptered California SB 53 text, Cal. Bus. & Prof. Code Chapter 25.1, and the California Attorney General's SB 53 page. Retrieved July 27, 2026._