Direct comparison
California AI Transparency Act vs SB 53
SB 942/AB 853 require AI-content labeling; SB 53 requires frontier-model safety reporting. Who each law covers, and which applies to you.
Written and maintained by CASRAI Editorial Board
Last updated
Ask CASRAI · free to try
Ask about California AI Transparency Act vs SB 53
Ask your first 2 questions free below. Subscribers get 150 a day for $29 a month.
Ask CASRAI answers research-administration questions and cites the passages behind every claim. When our sources don't cover a question, it says so.
Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.
Works on this site and inside Claude, Cursor and the AI tools you already use.
Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays free to read, with no account and no card.
How do California AI Transparency Act (SB 942 / AB 853), SB 53 (TFAIA) compare side by side?
The table below compares California AI Transparency Act (SB 942 / AB 853), SB 53 (TFAIA) across 7 procurement-relevant dimensions, from formal citation through enforcement & penalties.
Side-by-side comparison
| Dimension | California AI Transparency Act (SB 942 / AB 853) | SB 53 (TFAIA) |
|---|---|---|
| Formal citation | California AI Transparency Act, Cal. Bus. & Prof. Code §22757 et seq., as amended by AB 853 (2025) | Transparency in Frontier Artificial Intelligence Act (TFAIA), Cal. Bus. & Prof. Code §§22757.11–22757.15 |
| What it actually governs | Disclosure of AI-generated content — labeling and provenance metadata for images, video, audio and text produced by generative AI | Catastrophic-risk safety reporting for the largest frontier AI models — what a lab must publish about how it manages risk, and how fast it must report a serious incident |
| Who must comply | "Covered providers" — generative-AI systems with over 1,000,000 monthly visitors or users, publicly accessible in California. AB 853 extends separate duties to large online platforms, GenAI hosting platforms, and capture-device manufacturers. | "Frontier developers" (trained a model using more than 10^26 computing operations) and, with heavier duties, "large frontier developers" (a frontier developer whose group revenue exceeded $500 million in the prior year) |
| Core obligations | Mandatory latent (hidden metadata) disclosure embedding provider name, system version, and timestamp; optional manifest (visible) labeling; a free, publicly accessible AI-detection tool | A published "frontier AI framework" describing catastrophic-risk management; a transparency report before deploying a new/modified frontier model; a summary of internal risk assessments every three months |
| Incident / violation reporting | No safety-incident reporting duty. Licensees must maintain disclosure capability or face license revocation within 96 hours of a violation being discovered. | A "critical safety incident" must be reported to the California Office of Emergency Services within 15 days of discovery, or 24 hours if there is imminent risk of death or serious physical injury |
| Effective / compliance dates | Originally January 1, 2026; AB 853 delayed the core provisions to August 2, 2026. Large-platform and GenAI-hosting duties begin January 1, 2027; capture-device-manufacturer duties begin January 1, 2028. | Signed September 29, 2025; core obligations took effect January 1, 2026 |
| Enforcement & penalties | Civil penalties of $5,000 per violation, with attorney's fees recoverable by the enforcing agency | California Attorney General only — no private right of action. Civil penalties up to $1,000,000 per violation. |
Common questions
Common questions about California AI Transparency Act (SB 942 / AB 853) vs SB 53 (TFAIA)
My product is a public chatbot with over a million monthly users — which law applies to me?
+
Most likely the California AI Transparency Act (SB 942/AB 853), not SB 53. SB 942/AB 853 is triggered by being a consumer-facing generative-AI provider above the visitor threshold; SB 53 is triggered by training a frontier-scale model and, for the heaviest duties, by group revenue over $500 million. A company can be covered by SB 942/AB 853 without coming anywhere near SB 53's compute or revenue thresholds.
Does SB 53 replace SB 942/AB 853, or do both apply?
+
They are independent laws and can both apply to the same company. A large frontier-model developer that also ships a public-facing generative-AI product needs to check its obligations under both: SB 53 for the model's safety-framework and incident-reporting duties, SB 942/AB 853 for that product's content-labeling and detection-tool duties.
What did AB 853 change about the original SB 942?
+
AB 853 (2025) delayed SB 942's main compliance date from January 1, 2026 to August 2, 2026, and added new, later-effective duties for large online platforms and GenAI-hosting platforms (January 1, 2027) and capture-device manufacturers (January 1, 2028), on top of SB 942's original provider-facing disclosure requirements.
These laws all use words like "disclosure" and "transparency" differently — is there a way to keep the terminology straight?
+
That confusion is exactly what CASRAI's own NIKOLAI project exists to track. NIKOLAI is CASRAI's independent, unendorsed dictionary of frontier-AI-safety terminology — not an official standard and not endorsed by any lab, regulator, or evaluator — and its Transparency and Review track (N8) documents how terms like "disclosure," "transparency report," and "review" are used, and used differently, across frameworks including SB 53's frontier-model reporting duties. It won't tell you which law applies to your product, but it's a useful reference for spotting when two laws use the same word for two different obligations.
