Primary Research · State Profile · CA
California has the most fragmented and prolific AI regulatory landscape of any US state. Rather than a single comprehensive Act, California layers multiple AI laws on top of an already aggressive consumer privacy regime (CCPA/CPRA) and adds sector-specific overlays for chatbots, training data, automated decisionmaking, and frontier model safety. The 2026 portfolio includes AB 2013 (AI training data transparency, effective January 2026), SB 1001 (the 2018 'B.O.T. Act' bot disclosure rule), the CCPA/CPRA Automated Decisionmaking Technology regulations finalized in 2024, and CA Bar Practical Guidance for Use of Generative AI that bound lawyer AI use beginning November 2023.
California's regulatory posture in 2026 emphasizes transparency and process rather than prohibition. AB 2013 requires generative AI developers to publish detailed dataset disclosures on their public-facing websites by January 1, 2026 — covering whether training data includes personal information, intellectual property, dataset time ranges, and known limitations. Enforcement is by the California Attorney General with civil penalties up to $5,000 per violation. The CPPA's ADMT regulations layer on top, giving California consumers rights to access, opt out of, and request human review of automated decisions in employment, housing, insurance, and similar contexts. Organizations that already operate against the CCPA framework inherit much of the operational infrastructure needed for ADMT compliance, but the AI-specific notice and explanation obligations are genuinely new work.
Key provisions
Public website disclosure of training dataset composition (personal information, IP, aggregated data); disclose dataset time range, known limitations, and any modifications; up to $5,000 per violation; no private right of action.
Key provisions
Any bot communicating with Californians for commercial sale or to influence elections must disclose it is a bot — clear, conspicuous, reasonably designed to inform.
Key provisions
Pre-use ADMT notice; right to access information about ADMT use; right to opt out of automated decisionmaking; right to request human review for significant decisions (employment, housing, financial services, healthcare).
After Governor Newsom vetoed SB 1047 in September 2024, advocates have signaled reintroduction with narrower scope. The vetoed version would have imposed safety testing, kill-switch, and incident-reporting requirements on frontier AI developers training models above a defined compute threshold. Watch the 2026 session for narrower follow-on legislation.
Would have created an explicit cause of action for algorithmic discrimination in consequential decisions. Held in 2024 over concerns about overlap with existing FEHA and ADA frameworks; may return.
Sector-specific frameworks layer on top of state AI laws and frequently impose stricter or earlier-binding obligations. These are the sectors most exposed in California.
Legal
California Practical Guidance for the Use of Generative AI in the Practice of Law (November 2023) bound CA lawyers to specific AI-use disclosures, confidentiality controls, and competence obligations.
Healthcare
California AB 3030 (effective 2025) requires healthcare providers to disclose generative AI use in patient clinical communications and provide a means to contact a human provider.
Financial services
California Department of Financial Protection and Innovation oversight under the California Consumer Financial Protection Law overlays on top of ADMT regulations for credit and lending decisions.
Employment
California Civil Rights Council finalized regulations in 2025 prohibiting AI-driven employment discrimination; layers on top of ADMT.
Practical operational checklist for organizations subject to California AI laws. Items are ordered by typical sequence of implementation, not by importance — most steps depend on the inventory work in the first item.
Must be live by January 1, 2026 for any generative AI system made available to Californians.
Disclosure must be in the first conversational exchange and reasonably conspicuous.
Notice must precede the automated decision and explain logic in plain English.
Human review must be conducted by a qualified person with authority to reverse the decision.
AB 2013 disclosures require dataset provenance — most foundation model deployments need vendor cooperation to populate this.
Standard CCPA notices typically don't reach AI training and inference; explicit AI processing categories should be added.
SB 1047 successor bills are likely; large model developers should engage with legislative process early.
EFROS runs California AI governance as an integrated CCPA/CPRA + AI law program — AB 2013 dataset disclosure workflows, ADMT consumer notice templates, SB 1001 bot disclosure audits, and the CA Bar generative AI compliance overlay for law firm clients. We coordinate AI processing with existing CCPA infrastructure so privacy and AI compliance don't drift into separate, contradicting programs.
Disclaimer: this profile is a research dataset, not legal advice. Compliance determinations for California businesses require analysis of specific facts and should be made in consultation with qualified legal counsel licensed in California.
Reference this resource with attribution under CC-BY-4.0. Copy any of the formats below for academic papers, blog posts, AI citations, or vendor evidence packages.
Efros, S. (2026, May). California AI Law Tracker — 2026. EFROS. https://efros.com/research/state-ai-law-tracker/california/
Efros, Stefan. "California AI Law Tracker — 2026." EFROS, May 2026, https://efros.com/research/state-ai-law-tracker/california/.
Efros, Stefan. 2026. "California AI Law Tracker — 2026." EFROS. https://efros.com/research/state-ai-law-tracker/california/.
S. Efros, "California AI Law Tracker — 2026," EFROS, May 2026. [Online]. Available: https://efros.com/research/state-ai-law-tracker/california/
@misc{efros2026californiaailawt,
author = {Stefan Efros},
title = {California AI Law Tracker — 2026},
year = {2026},
month = {May},
publisher = {EFROS},
url = {https://efros.com/research/state-ai-law-tracker/california/},
note = {Accessed: May 2026}
}https://efros.com/research/state-ai-law-tracker/california/
Site-wide citation metadata is also published as a CITATION.cff file at /CITATION.cff for citation-management tools and academic indexers.
All tracked US state AI laws in a single citation-ready dataset.
Open20 AI vendors scored on 12 US AI governance axes — including state-law overlays.
OpenFederal framework that anchors most state AI law deployer obligations.
OpenOperating program that implements multi-state AI law compliance.
Open5-minute classification across Colorado AI Act, NYC LL144, CA AB 2013, and NIST AI RMF.
Open