California's New AI Laws: What Changes for Employers and Creators
On September 30, 2026, Governor Newsom signed 13 AI bills on workplace AI, watermarks, deepfakes and health care, and vetoed three. Here is who is affected.
By Yotam Dahan5 min readRegulation
California Governor Gavin Newsom signed 13 AI bills on September 30, 2026, adding rules for AI in firing decisions, workplace monitoring, health care, legal work, watermarks and deepfakes. On the same day he vetoed three other AI bills, including a ban on companion chatbots in therapy. Most of the new duties fall on employers, on platforms that handle AI content, and on professionals who use AI with clients or patients.
What the new laws do
The Governor's announcement groups the package by who it protects. It does not tie every rule to a single bill number, so treat this as a map and read the bill text before acting.
Firing and layoffs
Employers cannot rely only on AI for discipline or termination decisions, and must disclose when a mass layoff, relocation or termination is caused by an AI system.
Workplace monitoring
Employers may not use workplace surveillance tools in bathrooms.
AI content
Deleting a digital watermark is prohibited, metadata that shows whether content is AI-made becomes accessible, and public comments at local government meetings must come from real people.
Likeness and deepfakes
Civil and criminal likeness laws now cover AI deepfakes and impersonation, including explicit content created without consent.
Health care
Licensed providers keep their professional judgment when AI tools are involved, and makers of clinical decision tools must take reasonable steps to reduce known bias.
Lawyers
Attorneys may not fully hand core legal work, such as drafting briefs, to AI.
Here are the bills as the Governor's office listed them:
| Bill | Author | Topic |
|---|---|---|
| SB 947 | McNerney | Employment: automated decision systems |
| SB 951 | Gómez Reyes | Employment: notice of technological displacement |
| AB 1331 | Elhawary | Workplace surveillance |
| AB 1883 | Bryan | Workplace surveillance tools |
| AB 2713 | Wicks | AI Transparency Act: system provenance data |
| SB 1000 | Becker | AI Transparency Act |
| SB 1159 | Cabaldon | AI transparency and governance |
| SB 1111 | Ashby | Digital replicas |
| AB 1979 | Bonta | Health care services: AI |
| SB 503 | Weber Pierson | Health care services: AI |
| SB 574 | Umberg | AI and attorneys, arbitrators and judges |
| AB 2392 | Fong | AI procurement and training at public colleges |
| AB 1864 | Berman | Gene synthesis safety |
What Newsom vetoed, and why
The vetoes matter as much as the signatures, because they show where the line sits. According to the legislative update, three AI bills were returned unsigned:
- AB 2656 (public jobs). It would have required public employers to give unions at least 45 days' written notice before using generative AI for work their members do. The veto message says it would add redundant layers and that collective bargaining is the better route.
- SB 903 (therapy). It would have banned companion chatbots in psychotherapy and restricted therapists' use of AI. Newsom called it overly broad, saying it could capture general-purpose AI systems.
- AB 2575 (health care services). A third health care AI bill was also vetoed; its veto message is linked from the update.
The bigger picture is a state building an oversight system piece by piece:
- Earlier in September 2026SB 813 sets up certification for independent AI verification groups, and AB 1405 creates a state registry of AI auditors
- September 18, 2026An executive order speeds up independent audits and asks experts for a guide to stronger AI safety laws within two months
- September 30, 202613 AI bills signed and 3 vetoed
- Mid-November 2026The two-month deadline for the experts' guide, which may propose safety plans and an emergency shutoff for frontier models
What to do if you employ people or publish AI content
You do not need to panic, but a short audit now saves trouble later. If you have staff or customers in California:
0 of 6 done
An AI assistant can help you build the first draft of that inventory:
I run [TYPE OF BUSINESS] with [NUMBER] employees, some in California.
Help me list every place we use AI or automated tools, grouped as: hiring and HR, employee monitoring, customer communication, marketing images and video, and client advice.
For each one, ask me who reviews the output and whether a person makes the final decision.
Finish with a table of the items a lawyer should review first. Do not give legal conclusions.If you are building automations that touch staff or customers, our guide to AI automation for small business covers how to keep a human in the loop where it counts.
What to watch next
- The experts' guide. Due about two months after the September 18 order, it could become next year's frontier AI bills.
- Effective dates. Check each bill's final text; most California laws start on January 1 of the following year.
- Other governments. The EU's AI Act transparency rules for deepfakes and AI-generated content have applied since August 2, 2026. We track both on our AI news page.
Sources
Frequently asked questions
When do California's new AI laws take effect?
The Governor's announcement did not list dates. Under California's constitution, bills passed in a regular session normally take effect on January 1 of the following year unless the bill sets another date, so check each bill's text for its operative date.
Do California's AI laws apply to businesses outside California?
They can, depending on the law and on where your staff and customers are. If you employ people in California or publish content aimed at Californians, ask a lawyer how each new law applies to you.
Did California ban AI therapy chatbots?
No. The Legislature passed SB 903, which would have barred companion chatbots in psychotherapy, but Governor Newsom vetoed it on September 30, calling it overly broad.
