California Just Banned the Robo Boss: What the New Workplace AI Laws Actually Do

On September 30, 2026, California Governor Gavin Newsom signed a package of laws that makes California the first state to ban employers from letting an algorithm fire or discipline a worker on its own. The headline measure, the No Robo Bosses Act, was one of more than two dozen AI-related bills Newsom signed that day. The package also forces layoff notices to say when AI caused the cuts and bans workplace surveillance tools that try to read employees' emotions or collect their neural data. With no comprehensive federal AI law in sight, California is once again writing rules that every other state will be asked to read.

The No Robo Bosses Act: a human has to sign off

Senate Bill 947, the No Robo Bosses Act, was authored by state Sen. Jerry McNerney and backed by the California Federation of Labor Unions, AFL-CIO. It takes effect July 1, 2027, and bars California employers from relying solely on an automated decision system to discipline or terminate a current employee. If the employer primarily relies on the system's output, a human must corroborate the decision using the underlying data or supporting information such as personnel files. Output that cannot be corroborated, or that the reviewer finds inaccurate, incomplete, or misleading, cannot be used.

The law also creates notice rights. A worker disciplined or fired primarily on the basis of an automated recommendation must receive a separate, plain-language written notice, in the language ordinarily used for employee communications, stating the system's role, that a human reviewed and corroborated the decision, and giving contact information for a human the worker can ask. The employee can request a description of their data the system used, and the employer may not retaliate for exercising these rights. The law also bars using these systems to infer an employee's protected characteristics or to punish workers for exercising their legal rights.

The political backstory matters. McNerney tried a version of this bill in 2025 and Newsom vetoed it. The 2026 rewrite dropped an advance-notice requirement and narrowed the scope to exclude gig workers, and that was enough to earn the governor's signature. Unions pushed the bill as a response to algorithmic management, where workers described being timed, scored, and flagged by software with no human able to explain why.

The Labor Commissioner and public prosecutors enforce it through citations and civil actions, with penalties up to $500 per violation plus injunctions, punitive damages, and attorneys' fees. The employer bears the burden of showing it did not primarily rely on an automated system.

California has decided that firing someone is still a human decision. An algorithm can recommend it, but it cannot make it.

Layoff notices must now say "technology displacement"

A second law, SB 951, targets the other end of the employment lifecycle. Effective January 1, 2027, when AI or automation causes all or a substantial part of a covered mass layoff, relocation, or termination, the employer must add specified information to the notices required under Cal/WARN, the state law that generally requires 60 days' advance notice of mass layoffs. The notice must carry the statement "This notice is for a technology displacement" at the top, and must list the affected employees' numbers, classifications, and locations, the job functions being automated, and the category of AI or automated technology involved.

The state's Employment Development Department must also deliver a one-time report to the Legislature by January 1, 2028, on AI's effects on hiring practices. This is the first law of its kind to make AI-driven job loss a matter of public record at the moment it happens.

The surveillance bans: no emotion reading, no bathroom cameras

Two companion bills take aim at AI-enabled workplace surveillance. AB 1883, effective January 1, 2027, prohibits employers from using AI-enabled surveillance tools to infer or predict an employee's emotional state or to collect neural data by measuring the central or peripheral nervous system. The covered tools are defined broadly: systems that collect employee data, activities, communications, biometrics, or behavior without direct human observation, including video and audio surveillance, time tracking, geolocation, and electromagnetic tracking.

The law reaches private employers, governmental bodies, the University of California, and California State University. External applicants are not covered, leaving pre-employment recruiting tools outside the law. Limited exceptions exist for specified safety, aircraft, national security, military, space, and defense purposes. The Labor Commissioner and public prosecutors enforce it with the same remedies as SB 947.

AB 1331 bans surveillance tools in workplace bathrooms outright. Under the new regime, an employer using such tools in California must document any claimed statutory exception and disable, modify, or discontinue the prohibited function before January 1, 2027.

A voluntary audit regime, and lawyers too

The workplace laws were part of a much bigger package. An O'Melveny analysis published October 8, 2026, notes Newsom signed more than two dozen AI-related measures on September 30. Among them, SB 813 and AB 1405 create a voluntary framework for auditing and verifying AI risks, with a licensing regime for registered auditors. Audits are not mandatory, but undergoing one is described as "relevant to, but not conclusive of," liability in litigation over AI-caused harm.

SB 574 separately regulates attorneys' use of generative AI, and Newsom has ordered the state to accelerate implementation and study even tougher rules, including an emergency off switch for the most powerful AI models. California is not inching toward AI regulation. It is building the machinery all at once.

The rubber-stamp problem

The No Robo Bosses Act closes the most extreme scenario: an algorithm that fires a worker with no human involvement. But a legal analysis published October 5, 2026, argues the law requires a person to verify an AI-driven decision without specifying how much time the reviewer gets, whether they see the full data the algorithm weighed, or whether overriding the recommendation carries any professional cost for the reviewer. These are the same gaps that have produced rubber-stamping in prior human-oversight mandates, from content moderation to predictive policing: the reviewer exists on paper, approves nearly everything, and the algorithm effectively still decides.

The critique does not make the law meaningless: shifting the burden of proof to the employer and creating a paper trail of notices are substantive changes. But SB 947 reads as a first draft of a governance problem no jurisdiction has fully solved. That is analysis, not a claim about the law's text.

Why California's rules will travel

A Seyfarth legal update published October 9, 2026, notes the new statutes sit apart from the California Consumer Privacy Act's automated decisionmaking technology rules and reach employers and public entities the CCPA does not cover. That breadth matters because other states are already moving. Colorado, Illinois, and Texas have recently passed narrower AI workplace regulations, and Connecticut's AI and privacy laws took effect October 1, 2026. One political assessment published in October 2026 argues these laws will become a model other states copy, precisely because limiting robo bosses is politically popular.

For national employers, the compliance map just got more complicated. For workers elsewhere, California's experiment is a preview of what their own legislatures may soon be asked to vote on.

This article explains reported policy developments and is not legal advice.