Transparency in Frontier Artificial Intelligence Act (SB 53)

TFAIA

Artificial intelligence

An AI safety transparency law rather than a privacy law: large frontier AI developers must publish a framework for managing catastrophic risk, publish transparency reports when releasing frontier models, and report critical safety incidents to the Office of Emergency Services. Included here because it is one of California's core AI statutes.

Where: California
Citation: Cal. Bus. & Prof. Code 22757.10-22757.16
Status: In force
In force since: 2026-01-01
Enforced by: California Attorney General (civil penalties); Office of Emergency Services (incident reports)
People can sue: No
Penalties: Civil penalty up to $1,000,000 per violation, by severity, in AG actions only (22757.15).
Applies to: Frontier developers: those that trained a foundation model using more than 10^26 operations
Large frontier developers (frontier developers above a revenue threshold) for framework and risk-reporting duties

What a privacy notice must say

Large frontier developers must write, follow, and publish a frontier AI framework and review it yearly.Cal. Bus. & Prof. Code 22757.12(a)-(b)
Publish a transparency report when deploying a new or substantially modified frontier model.Cal. Bus. & Prof. Code 22757.12(c)

Practices it requires

Do not make materially false or misleading statements about catastrophic risk or framework compliance.Cal. Bus. & Prof. Code 22757.12(e)

Other duties

Report critical safety incidents to the Office of Emergency Services within 15 days (24 hours if imminent risk of death or serious injury).Cal. Bus. & Prof. Code 22757.13(c)

Sources

Checked against these sources on 2026-09-25 by research agent (Claude), primary sources.

Unverified: Revenue threshold for 'large frontier developer' ($500 million) and whistleblower provisions (Labor Code 1107 et seq.) are from background knowledge. | Last section number of the chapter (22757.16) not confirmed.

Research reference, not legal advice.