California Sends Newsom AI Bills on Robo-Firing, Layoff Notices and Audits

California's legislature closed its 2026 session late on Monday, August 31, after passing about 30 AI-related bills, according to the Transparency Coalition, which tracks AI legislation. Two were already signed and the rest went to Gov. Gavin Newsom, who has until September 30 to sign or veto them. Beyond chatbots and children's safety, the package contains detailed rules on how employers use AI to manage people, plus new machinery for auditing AI systems. For any company with workers or customers in California, this is the week to start mapping where those rules would land.

What did California's legislature pass on AI?

The bills fall into a few groups. The ones most relevant to large enterprises are:

• Automated decisions about workers. SB 947, from Sen. Jerry McNerney with Sen. Eloise Gómez Reyes as a coauthor, won final approval on August 31, Bloomberg Law reported. It would bar employers from relying solely on an automated decision system for discipline or firing. When an employer primarily relies on one, a human reviewer would have to investigate and corroborate the output, and the worker would get written notice, Allwork.Space reported. The bill would take effect July 1, 2027, if signed, with enforcement by the Labor Commissioner and public prosecutors. It is a revised version of a 2025 bill that Newsom vetoed as overly broad, The Sacramento Bee reported.
• Notice of AI-driven layoffs. SB 951, from Sen. Reyes, would require covered employers to give advance notice before technological displacement affecting 25% or more of a workforce. The Transparency Coalition summary describes a 90-day notice period. Law firm Wiley said the bill also adds state reporting requirements when layoffs substantially result from AI or automation.
• Workplace surveillance. AB 1883, from Asm. Isaac Bryan, would bar AI-enabled monitoring that collects workers' neural data or infers their emotional state.
• Audits and certification. AB 1405 would create a state registry of AI auditors and bar unregistered firms from conducting AI audits. SB 813 would set rules for independent verification organizations that certify AI systems for safety and compliance.
• Customer-facing AI and provenance. AB 1609 sets disclosure and operating rules for customer service chatbots at large businesses, while SB 1000 and AB 2713 expand the state's AI Transparency Act rules on disclosure and provenance data.

What are supporters and opponents arguing?

The workplace bills drew the sharpest debate. McNerney has framed SB 947 as an oversight measure rather than a ban, telling Capitol Weekly in July that lawmakers want these systems used "with human oversight." The California Labor Federation sponsored several of the employment bills and has argued that the state needs accurate data on AI-related job losses.

The California Chamber of Commerce opposed SB 947, saying it would raise costs and compliance burdens and invite litigation, and called SB 951 premature given a governor's executive order that directed a review of worker protections, according to the same Capitol Weekly report. Restaurant, retail and grocery groups also opposed SB 947 at final passage, the Bee reported. Local government groups argued some bills applied private-sector rules poorly to public employers. Several bills were amended as they moved, so the final text matters more than earlier summaries. Early versions of SB 951, for example, called for 60 days' notice.

Newsom's decisions are the open question. Until he acts, none of these are law.

Does this change how enterprises should govern AI in HR?

In practical terms, SB 947 would turn a common governance principle into a legal requirement for one high-stakes use case. Many enterprises already say a human makes the final call on discipline and termination. The bill would require them to prove it: a documented independent review, corroborating evidence, a notice to the employee and, on request, a description of the data used.

That is the same evidence problem AI Connect speakers have raised for agentic systems generally. Reggie Townsend of SAS put it this way at The AI Enterprise Conference 2026: "Trust is a feeling. Trustworthiness is the evidence," as we reported in SAS on closing the AI trust gap. In a AI Connect expert panel in August, Shubham Trivedi, Digital Solution Advisor at SAP, argued that oversight should scale with risk: "The goal isn't to keep a human in the loop. It is to make sure the human is a step away from the loop where it matters." Firing decisions are a clear case where the human needs to be in the loop itself, and our coverage of where humans belong in agentic AI governance offers a risk-tiering model that maps well to these rules.

There is a reasonable counterargument. Human corroboration adds cost and time, and a rushed reviewer who rubber-stamps model output satisfies the form of the rule without its intent. Employers using automated scheduling, productivity scoring or deactivation tools may find the definitions broad.

What should enterprise leaders do now?

Even before Newsom acts, the package points to a direction other states may follow. Four steps make sense regardless of the outcome:

1. Inventory AI used in people decisions. List every system that scores, ranks, flags or schedules employees and contractors, including vendor tools embedded in HR and workforce platforms.
2. Document the human review. Where a model informs discipline or termination, record who reviewed it, what other evidence they used and how the worker was told.
3. Plan for displacement disclosure. If automation plans could affect a large share of a site or function, decide now how you would describe and report that, rather than waiting for a deadline.
4. Treat audits as a market that is about to be regulated. If AB 1405 and SB 813 become law, third-party AI assurance in California would need registered providers, which affects which firms you hire for model reviews.

AI Connect practitioners have long argued that governance built into the workflow speeds delivery, as we covered in governance by design. Companies that already built review, logging and notice into their AI workflows will find compliance mostly a paperwork exercise. The rest have until the governor's September 30 deadline to see what they will be building toward.

Frequently asked questions

What is California SB 947?

SB 947, the No Robo Bosses Act, would bar employers from relying solely on automated decision systems for discipline or termination, require human corroboration and give workers notice. It would take effect July 1, 2027, if signed.

When does Newsom have to act on California's 2026 AI bills?

Gov. Gavin Newsom has until September 30, 2026, to sign or veto the bills the legislature passed before adjourning on August 31.

What should employers do before the bills become law?

Inventory AI tools used in people decisions, document human review of model outputs, plan how to disclose automation-driven displacement and review which firms conduct AI audits.

Filed under: AI Governance & Trust, People, Culture & Leadership