No Robo Bosses in the Golden State: California Outlaws AI-Only Firings and Unreported AI Layoffs

On September 30, 2026, California Governor Gavin Newsom signed a package of first-in-the-nation worker protections that directly target the way artificial intelligence is deployed against employees. At the center of the package is SB 947, the “No Robo Bosses Act of 2026” — the first law in any U.S. state that flatly prohibits an employer from letting an automated decision system (ADS) fire or discipline a worker on its own. Alongside it came SB 951, which amends the California WARN Act to force employers to disclose when AI is the driving force behind a mass layoff, and a new surveillance restriction banning AI tools that infer workers’ emotions or neural signals. With the ink now dry, employment lawyers across the country are calling it the most consequential workplace-AI regime in the United States — and the compliance clock is already running.

What SB 947 actually does

The No Robo Bosses Act, authored by Senator Jerry McNerney, takes aim at a practice that has quietly spread through logistics, retail, and customer-support operations: fully automated discipline and termination. Under the bill, California employers are barred from relying solely on an automated decision system to make disciplinary or termination decisions. When an employer “primarily relies” on an ADS for such a decision, the affected employee must be notified and given a meaningful right to human review — a person, not a pipeline, has to look at the outcome before it sticks.

Crucially, the law goes further than a simple human-in-the-loop mandate. It also prohibits certain uses of automated systems regardless of whether a human signs off at the end — most notably, systems that make or act on inferences about protected characteristics. If a productivity-scoring model quietly penalizes workers for inferred pregnancy, union sympathy, or disability, no amount of nominal human oversight launders the practice. Those inferences themselves become unlawful.

There is a notable political footnote: Newsom vetoed a broader version of the robo-boss legislation last year, arguing it would overreach. SB 947 is the narrower rewrite that survived — a pattern that has defined his approach to AI regulation all session. The law takes effect July 1, 2027, giving employers roughly nine months to audit their systems.

SB 951: AI-driven layoffs can no longer be silent

The second pillar, SB 951, amends the California Worker Adjustment and Retraining (Cal/WARN) Act — the statute that already requires 60 days’ advance written notice before a mass layoff, relocation, or termination. Under the amendment, whenever AI is a cause of such an event, the standard notice must be expanded with additional AI-specific disclosures: what role the technology played, which kinds of jobs are affected, and related details that give workers and state agencies a real picture of automation’s footprint in the decision.

The practical effect is significant. Companies have grown fond of describing AI-driven restructuring as ordinary “efficiency” or “reorganization.” In California, if a recommendation engine, workforce-optimization platform, or agentic system is what triggers the cuts, that fact now has to appear in the paperwork that laid-off workers actually receive. Fisher Phillips and other employment firms are already advising clients to fold four new pieces of AI-related information into their WARN notices.

The surveillance ban: no emotion tracking, no neural data

The package also confronts workplace monitoring. A companion measure prohibits employers from deploying AI-powered surveillance tools capable of recognizing — or making inferences and predictions about — workers’ emotional states or neural data. The restriction covers tools built on video, audio, time-tracking, geolocation, and even electromagnetic-signal monitoring, and it separately bars surveillance of employees in workplace bathrooms. For an industry that has marketed “affect detection” and “attention scoring” to warehouse and call-center operators, California just closed the market.

Context: the end of a 30-bill season

The signing caps a legislative sprint in which California lawmakers pushed roughly 30 AI bills to the governor’s desk, covering everything from chatbot risk assessments to an AI kill-switch executive order. Newsom ultimately signed thirteen AI bills in this final wave, including the No Robo Bosses Act, while vetoing a smart-glasses privacy bill he considered too broad. The state now has, by its own account, “the nation’s leading AI framework” — a claim that lands harder given the federal posture. While Washington leans on voluntary accords and a White House executive order literally renaming AI as “Super Intelligence,” Sacramento is writing enforceable rules with dates, penalties, and private rights of action attached.

Other states have nibbled at the edges — Illinois regulates AI analysis of video interviews, New York City requires bias audits for automated hiring tools, Colorado governs high-risk AI generally. None had touched termination decisions until now. California is the fifth-largest economy in the world; where it goes on employment law, multistate employers tend to follow.

What employers should do before July 2027

Legal analyses published in the first days of October converge on a short list of actions. First, inventory every automated decision system that touches hiring, scheduling, productivity scoring, discipline, or termination — including those buried inside vendor platforms. Second, stand up a human-review workflow with real authority to override ADS outcomes, and document it. Third, rework Cal/WARN notice templates to accommodate the new AI disclosures. Fourth, scrub any analytics that infer protected attributes or emotional states, because those uses are unlawful regardless of oversight. And fifth, revisit vendor contracts: if a supplier’s platform cannot support notice and review obligations, the platform is a liability.

Why it matters beyond California

The deeper significance is architectural. For two years, enterprise AI adoption has been sold on the promise of autonomy — agents that act, decide, and optimize without waiting for a human. California has now drawn a bright line: in the employment relationship, autonomy has a ceiling. An AI can recommend, rank, and draft, but the decision to end someone’s livelihood belongs to a person who can be held accountable. As agentic systems spread into HR stacks, expect this principle — “recommendation, not termination” — to become the template that other states copy and that AI vendors must design around. The robo boss didn’t just lose its job in California. It lost its job description.