- within Technology topic(s)
- with readers working within the Technology, Media & Information and Oil & Gas industries
California's new Transparency in Frontier Artificial Intelligence Act creates the nation's first standardized safety disclosure framework for frontier AI models.
On September 29, 2025, Governor Newsom signed the Transparency in Frontier Artificial Intelligence Act ("TFAIA"), establishing the nation's first standardized safety disclosure regime for "frontier" AI model developers. While less stringent than its failed predecessor, SB 1047, TFAIA imposes a host of obligations to facilitate public safety, including transparency requirements, reporting duties, and whistleblower and employee protections.
Scope
TFAIA applies to developers of "frontier" AI models—models trained using over 10^26 computational operations—and "large frontier developers" ("LFDs")—those with annual gross revenues exceeding $500 million (collectively, "Developers").
Transparency
- Developers must publish "transparency reports" describing their AI models, including intended uses, restrictions, and contact mechanisms. LFDs must also include an assessment of "catastrophic risks"—certain risks that materially contribute to death or serious injury of over 50 individuals or over $1 billion in damages or property loss.
- LFDs must also publish a "frontier AI framework," outlining technical and organizational protocols to manage, assess, and mitigate catastrophic risks, including, among others, descriptions of how LFDs incorporate standards and best practices, and thresholds set to assess catastrophic risks.
Reports to Office
- Developers must report "critical safety incidents"—those relating to death, bodily injury, property damage, or model control loss—to the Office of Emergency Services ("OES") within 15 days of discovery, or to an applicable authority within 24 hours if posing an imminent risk of death or serious injury.
- LFDs must also submit to the OES periodic summaries of potential catastrophic risk assessments from internal use of their frontier models.
Employees & Whistleblowers
- TFAIA provides whistleblower protections for employees responsible for assessing, managing, or addressing risks of a model's critical safety incidents.
- Developers must notify such employees regarding their rights under TFAIA, while LFDs must also implement internal processes for those employees and high-level personnel to report and receive updates on risks and alleged violations of TFAIA.
Enforcement
- LFDs face steep civil penalties of up to $1,000,000 per violation including for non-compliance with transparency, reporting, or framework requirements, enforced by the Attorney General.
While TFAIA focuses on Developers, it also impacts deployers and the industry more broadly. Developers may consider implementing policies and procedures to establish transparency, reporting, and framework requirements, alongside employee trainings and internal reporting processes regarding critical safety incidents and catastrophic risks. Deployers may consider reviewing vendor relationships, establishing policies and procedures to review transparency reports and frontier AI frameworks, and/or incorporating contractual provisions requiring Developers to provide such documents.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
