Newsom's Sept 30 AI Signings: Who Must Act

TL;DR Newsom signed a large batch of AI bills on September 30, 2026. The ones most likely to touch a small team are SB 947 (operative July 1, 2027), SB 951 (operative January 1, 2027), SB 574 (lawyers and arbitrators), AB 1979 and SB 503 (clinical AI), and SB 1000 and AB 2713 (AI Transparency Act). This page lists who each bill reaches, what it asks for, and which dates we could verify.

The governor's office published the list on September 30, the last day he had to act on bills from the 2026 session. The release names the bills and gives a one-line description of each, but no effective dates. We went to the bill text and to law firm and press coverage for the dates and duties, and we mark below where a date is confirmed and where it is not.

Two earlier pieces on this site predicted outcomes before the signing: the August 19 SB 947 and SB 951 article and the September tracker of California's 2026 AI bills. Where this page and those differ, this page is newer and is based on the signed text.

The signed bills at a glance

The governor's release also lists AB 1331 and AB 1883 (workplace surveillance tools in bathrooms), AB 2392 (generative AI procurement and training in public postsecondary education) and AB 1864 (gene synthesis equipment providers). We did not read those texts, so they are outside the analysis below.

Employers: SB 947 and SB 951

SB 947 sets limits on how an employer uses an automated decision system, or ADS, in discipline and termination. The bill defines an ADS broadly: a computational process derived from machine learning, statistical modeling, data analytics or AI that issues an output such as a score, classification or recommendation used to assist or replace human decision making and that materially affects a person. Spam filters, firewalls and similar tools are excluded.

What the signed text asks of an employer, as we read it:

• No sole reliance. An employer cannot rely solely on an ADS for a disciplinary or termination decision.
• Human corroboration. When an ADS plays a primary role, a human must corroborate the output. If the corroboration fails or shows the output was inaccurate, the output cannot be used to justify the action.
• Written notice. The employee gets a notice saying that an ADS primarily drove the decision, that a human corroborated it, who to contact with questions, and that retaliation is prohibited.
• Prohibited uses. An ADS cannot be used to violate labor, health and safety or civil rights law, to infer an employee's protected status, or to predict and retaliate against workers who exercise legal rights.
• Enforcement. The Labor Commissioner and public prosecutors can enforce. The civil penalty is $500 per violation, with injunctive relief, punitive damages and attorney fees available in civil actions.

The operative date is July 1, 2027, which is six months later than the January 1, 2027 date our August article guessed. Two independent sources agree on July 1.

SB 951 does not create a new notice. It amends Cal/WARN so that when a mass layoff, relocation or termination is caused in whole or in substantial part by an AI system or other automated technology replacing or automating jobs, the notice must say so. The notice must start with the line "This notice is for a technology displacement" and identify:

1. the number, classification or occupation, and work location of the affected workers,
2. the job functions that will be automated, and
3. the specific category or type of AI system or automated technology that substantially caused the displacement.

The Employment Development Department publishes summaries of these notices and quarterly statewide summaries of reported technology displacement. The bill is operative January 1, 2027.

Our August article, and several law firm alerts written before the final amendments, described a 90-day notice, a 25-worker threshold, a right of first bid and a dedicated penalty fund. The enrolled text we read uses Cal/WARN's existing 60-day notice and existing covered-establishment thresholds (at least 75 employees, and a layoff of at least 50 employees in a 30-day period), and we found no new penalty. Cal/WARN's existing remedies of up to $500 per day apply in our reading, but the bill itself does not spell out how they attach to the new disclosures, so ask counsel before relying on that.

Lawyers and legal teams: SB 574

SB 574 is the first state statute written around how lawyers and arbitrators use generative AI. As described in the bill text and in Bloomberg Law's report of the signing:

• An attorney cannot delegate the practice of law to generative AI.
• A brief, pleading, motion or other paper filed in court cannot contain a citation that the responsible attorney has not personally verified, including citations produced by AI.
• Attorneys must take reasonable steps to verify AI output, keep nonpublic client information out of AI systems that lack access restrictions, and correct erroneous content.
• Arbitrators cannot delegate any part of their decision making to generative AI and must disclose AI-generated information from outside the record to the parties.

If you buy or build AI tools for a legal team, the contract question is whether the tool shows its sources in a way an attorney can verify, and whether it keeps prompts out of any shared model.

Health care: AB 1979 and SB 503

AB 1979 bars health facilities, clinics and physician offices from using an AI tool to independently perform a clinical function that the law requires a licensed professional to perform. It also requires reasonable steps to ensure that a licensed provider keeps independent professional judgment over decisions that use a clinical decision support system. Uses such as risk scoring, alerts and clinical summarization remain allowed when a licensed professional keeps decision authority. The Medical Board of California's analysis gives a January 1, 2027 effective date.

SB 503 applies to developers and deployers of AI clinical decision support systems. Developers must make reasonable efforts to identify known or reasonably foreseeable risks of biased impacts, mitigate them, and give deployers documentation: intended uses, risk assessments, training data summaries and monitoring recommendations. Deployers must monitor the systems and take reasonable steps to mitigate bias. The text we read names no dedicated penalty, and it says compliance is not a defense to a discrimination claim.

AI providers and platforms: SB 1000 and AB 2713

SB 1000 amends the California AI Transparency Act. Reported changes are the removal of the one million monthly user threshold from the covered provider definition, retention of the machine-readable provenance disclosure, and removal of the visible disclosure requirement. It is described as an urgency measure effective on signing. If your product generates images, video or audio for people in California, read the chaptered text now rather than waiting for a January date.

AB 2713 requires large online platforms to provide an interface that indicates whether content was generated or substantially altered by a generative AI system, with a safe harbor for provenance data that does not follow widely adopted specifications.

A five-minute applicability screen

Paste this into your own tracker and answer each line. A "yes" means read the section above.

California Sept 30, 2026 AI signings: applicability screen
1. Do we have employees in California and use any tool that scores,
   flags or recommends in discipline or termination?   (SB 947, July 1, 2027)
2. Could a layoff, relocation or termination be caused by automating jobs?
   (SB 951, Jan 1, 2027)
3. Do our lawyers or arbitrators use generative AI for filings or decisions?
   (SB 574, check operative date)
4. Do we operate a health facility, clinic or practice that uses AI in
   clinical decisions?  (AB 1979, Jan 1, 2027)
5. Do we build or deploy clinical decision support AI?   (SB 503)
6. Do we offer a generative AI system available in California?   (SB 1000)
7. Are we a large online platform that shows user-generated content?   (AB 2713)
8. Do we use a person's voice or likeness in ads or media?   (SB 1111)
Owner: ______   Date reviewed: ______   Counsel consulted: yes / no

For SB 947 you can also pre-build the written notice. These are the fields the signed text calls for:

Notice to employee (SB 947)
- An automated decision system primarily drove this decision: yes
- A human corroborated the system output before action: yes (name, date)
- Questions: contact [name, email, phone]
- We do not retaliate against employees who ask about or contest this decision

For SB 951, the layoff notice needs the header line "This notice is for a technology displacement" plus the three disclosures listed above. Put the three fields in your standard Cal/WARN template now, left blank unless a layoff is automation-driven.

What we could not verify

• We read the bill texts through summaries produced by a tool and cross-checked the main points against law firm and press coverage. Dates for SB 947 and SB 951 are confirmed by two independent sources. Dates for SB 574, SB 503, AB 2713, SB 1111 and SB 1159 are not confirmed. California statutes that are not urgency measures normally take effect January 1 after enactment, but a bill can say otherwise.
• The governor's release gives no effective dates and no penalty amounts.
• We did not read AB 1331, AB 1883, AB 2392 or AB 1864.
• The Cal/WARN penalty point above is our reading, not a quote from the bill.

What to do this week

1. Add SB 951 fields to your Cal/WARN template before January 1.
2. Inventory scoring and flagging tools in your HR stack against the SB 947 definition, and list which of them feed discipline or termination.
3. If you sell or run generative AI to California users, have counsel read SB 1000 now.
4. Calendar July 1, 2027 for SB 947 and January 1, 2027 for SB 951 and AB 1979.

Related Reading

• California SB 947 and SB 951: Employer AI Rules
• California 2026 AI Bills Tracker
• California Adam's Law (SB 1119) Chatbot Compliance
• California ADMT Employer Compliance Templates
• Multi-State AI Compliance Strategy
• Newsom EO N-9-26: AI Kill Switch Contract Clauses