TL;DR: California's legislature passed 26 AI bills before adjourning August 31. Two are already signed. Twenty-four sit on Newsom's desk until the September 30 deadline. Based on his documented veto history, roughly 18 will likely become law. Most take effect January 1, 2027. The bills cover chatbot safety, deepfakes, employer surveillance, and AI auditor rules. This tracker maps what each category means for your team.
California's legislature ran until near midnight on August 31, 2026, passing 26 AI-related bills in the final hours before adjournment. Combined with 8 privacy bills passed in the same session, the output is one of the largest single-session AI legislative packages any U.S. state has produced.
Governor Gavin Newsom has until September 30, 2026 to sign or veto each one. After that date, unsigned bills die automatically. Compliance teams cannot wait for the October 1 morning-after summary -- the bills that will require systems changes need preparation now, not three weeks from now.
One bill is already law. AB 1651, which regulates attorney use of AI before the State Bar of California, was signed on August 22 before the session even closed. The other 24 are in the governor's queue.
Newsom's veto pattern: what the data says
Newsom has signed and vetoed enough AI bills at this point that his decision logic is readable. The pattern:
He signs: Transparency requirements. Disclosure mandates. Consumer protection rules. Child safety bills. Criminal law extensions. Bills that tell companies what to tell people.
He vetoes: Bills that restrict how AI can be developed. Bills that impose liability for AI outputs without a clear harm and intent threshold. Bills that a reasonable person could read as putting California at a competitive disadvantage for AI investment.
The clearest data point is the 2024 SB 1047 veto. That bill would have required frontier AI developers to conduct third-party safety evaluations and publish the results before training runs above a certain compute threshold. Newsom called it "well-intentioned but flawed" and said it would have driven AI investment out of California. He signed SB 53 in the same cycle -- a transparency and whistleblower bill that required frontier AI companies to publish safety frameworks and protect employees who raised safety concerns internally.
The read for 2026: bills that require disclosure, add safeguards for identifiable vulnerable populations, or extend existing criminal law to cover AI conduct are likely to be signed. Bills that restrict what AI products can do, or impose liability without a specific harm theory, are uncertain to unlikely.
Chatbot safety bills (3 bills -- likely to sign)
SB 867 amends California's companion chatbot law to prohibit manufacturing or selling companion chatbots embedded in toys. This is a narrow product-safety restriction on a specific category -- AI chatbots marketed to children in physical toy form. It does not restrict AI development broadly and has no plausible Big Tech opposition argument. Likely signed.
SB 1119 (Adam's Law) amends the same companion chatbot law to require operators of chatbots that interact with minors to submit to independent child safety audits. It also requires safety testing protocols before release. The law is named after Adam Raine. Child safety audit requirements are squarely in Newsom's signed category. Likely signed.
AB 1609 creates new disclosure obligations for customer service chatbots operated by large private businesses -- companies with more than $500 million in gross annual revenue nationally. The bill requires companies to disclose when a customer is interacting with an AI rather than a human and provide an option to escalate to a human agent. Disclosure plus a size threshold that limits impact on small businesses fits Newsom's profile. Likely signed.
What this means for compliance teams: companies above the AB 1609 revenue threshold need to audit their customer service AI stacks now. A September 30 signature means January 1, 2027 is the compliance deadline. Eight weeks is not much runway if you are operating AI customer service at scale.
Deepfake and synthetic media bills (3 bills -- likely to sign)
SB 1015 extends California's existing laws on contact with a minor to cover threats and extortion carried out using AI-generated deepfake imagery. This is a criminal law extension, not a new AI regulation -- it applies existing liability frameworks to new conduct enabled by AI. Newsom has consistently signed criminal law AI extensions. Likely signed.
SB 1111 clarifies that California's false impersonation statute covers the use of a digital replica when the intent is to defraud. Again, this is statutory clarification rather than a new regulatory framework. The bill closes a gap that existing law did not explicitly address. Likely signed.
SB 1050 requires disclosure when AI synthetic performers -- AI-generated likenesses of people -- appear in advertisements. Companies running AI-generated ad content must disclose that the performers are synthetic. This is a disclosure bill. Likely signed.
The practical impact of SB 1050 is broader than it sounds. If your marketing team is generating AI video or image content featuring synthetic human presenters, you need a disclosure protocol before January 1. That applies to brands using AI ad creation tools, not just the AI companies themselves.
Employer AI and worker surveillance (1 bill -- uncertain)
AB 1883 is the most uncertain bill in the batch. It prohibits an employer from using a workplace surveillance tool that uses AI to collect neural data or recognize an individual's emotional state. The prohibition covers tools that monitor workers and infer emotional status from facial expressions, voice tone, or other biometric signals.
Newsom has signed worker protection AI bills before -- SB 947 and SB 951 (the "no robo-bosses" bills from the prior session) both passed with his signature. But AB 1883 goes further by restricting a category of AI product capability rather than just requiring notice or appeal rights. Some workplace safety tools and HR analytics platforms use emotional state detection as a feature.
The bill's opponents will argue it restricts legitimate workplace safety monitoring. Its supporters will point to the specific category (neural data and emotional state) as the narrow target, not general employee monitoring. Newsom's decision will likely turn on whether he reads this as a privacy bill or an AI development restriction. Outcome: uncertain.
If AB 1883 is signed, HR tech vendors and enterprise customers using emotional AI features in their workforce tools need to plan for removal or modification by January 1. Start the vendor conversation now rather than waiting for the October outcome.
AI auditor registry (1 bill -- likely to sign)
AB 1405 directs the Government Operations Agency to establish an AI Auditor Registry. AI auditors would register with the state, and the registry would be publicly accessible. This creates an administrative infrastructure bill rather than imposing new compliance obligations on companies. Newsom has generally supported AI governance infrastructure -- he signed the AI Incident Reporting requirements in 2025. Likely signed.
What this means: if you are building an AI audit practice or hiring third-party AI auditors, the California registry will likely become a qualifying credential. Companies operating in California that use third-party AI audits may find that registry membership becomes a due diligence requirement.
What to prepare for now -- regardless of Newsom's decisions
The September 30 deadline is not when compliance starts. It is when the law is set. If you wait until October 1 to begin preparing, you have two to three months before laws take effect -- not enough time to modify customer-facing AI systems, retrain vendors, or update employment policies.
Run these audits in September:
-
Map every customer-facing AI system to the AB 1609 threshold. If your company is above $500 million in annual revenue and uses AI in customer service, the disclosure requirement likely applies.
-
Audit your marketing AI outputs for synthetic performers. Any AI-generated video or image content featuring human-like figures needs a disclosure protocol before January 1 if SB 1050 is signed.
-
Review your HR tech stack for emotional state detection features. If you use workforce analytics tools that infer employee emotional or mental states from behavioral signals, get a vendor answer on whether those features qualify as neural or emotional AI surveillance under AB 1883.
-
If you retain outside AI auditors, check whether your agreements require auditors to be registered with a California state registry. That requirement does not yet exist, but if AB 1405 is signed, it may become a de facto standard in California contracts within 12 months.
The deadline that matters is September 30 -- not because Newsom announces his decisions on that date, but because your planning needs to account for the realistic outcome of roughly 18 of 24 bills becoming law. Waiting for certainty means starting compliance work in October. For systems changes, eight weeks to January 1 is tight.
For related California AI law compliance, see the no robo-bosses worker protection guide, the AB 1898 workplace AI notice guide, and the AI regulation deadline calendar.
Related Reading
- California SB 947 / SB 951: no robo-bosses, worker AI protection
- California AB 1898: 90-day employer AI notice requirements
- California SB 942: AI Transparency Act enforcement (Midjourney)
- California SB 942: AI Transparency Act August 2026 compliance guide
- AI regulation deadline calendar 2026
- AI securities class actions 2026: disclosure checklist
- FTC AI enforcement actions 2026
