California Bans AI ‘Robo Bosses’: First-in-Nation Law Requires a Human Hand in Firing Decisions

The next time an algorithm decides a worker deserves to be fired in California, a human being will have to look that decision in the eye — and back it up with evidence. Governor Gavin Newsom on Wednesday signed SB 947, the “No Robo Bosses Act of 2026,” making California the first state in the nation to prohibit employers from relying solely on AI-powered automated decision systems to discipline or terminate employees.

The signing caps a remarkable reversal for the legislation. Newsom vetoed an earlier version of the bill just last year, even after it cleared both chambers of the state legislature with broad support, citing concerns that vague notice requirements could chill the adoption of beneficial workplace technologies. The version that reached his desk this time was narrowed to win his approval — and landed amid record public distrust of AI’s growing role in the workplace, according to a May survey by the nonprofit United for Respect, which found workers at major retailers increasingly worried that HR decisions are being automated.

What the Law Actually Requires

At its core, SB 947 establishes a “human-in-the-loop” mandate for the most consequential employment decisions. Employers cannot rely exclusively on an automated decision system when disciplining or firing a worker. Crucially, the law goes further than requiring a rubber stamp: a human reviewer must corroborate the machine’s recommendation using additional information such as managerial evaluations, peer reviews, or personnel records.

Transparency is the second pillar. Workers subject to such decisions must be told in writing that AI played a primary role, informed of what personal data the system drew upon, and given the name of a human contact who can walk them through the outcome. Enforcement falls to the California labor commissioner, the state attorney general, or local prosecutors. The law takes effect in July 2027, giving employers a runway to audit their HR technology stacks.

Not Just One Bill — A Whole Package

SB 947 was the centerpiece of a slate of thirteen AI-related bills Newsom signed the same day, an unprecedented regulatory sweep spanning employment, health care, education, and the legal profession. Other measures signed Wednesday include a ban on workplace tools that use AI to read a worker’s emotional state or collect neural data, a prohibition on monitoring employees in bathrooms with audio, video, or AI tools, and a requirement that layoff notices include additional information when jobs are being eliminated by technological displacement. In health care, AB 1979 prevents medical providers from using AI in ways that replace the clinical judgment of licensed professionals, and confidentiality protections were extended to health chatbots.

The political timing was pointed. Newsom signed the package one day after President Trump hosted executives from Google, Meta, Anthropic, OpenAI, xAI, and Nvidia at the White House, where they signed a voluntary AI safety agreement with no legal enforcement mechanism. Newsom dismissed the event bluntly, saying what he heard in Washington “should scare the hell out of everybody,” and positioned California’s approach — enforceable state law rather than voluntary pledges — as the model the country should follow.

The Battle Isn’t Over

Business groups fought the bill to the end. The California Chamber of Progress warned Newsom in a letter that the law’s trigger threshold — whether an employer “primarily relies” on an automated system — is never defined, creating compliance uncertainty. Opponents argue the ambiguity could discourage employers from using technologies that improve consistency and help managers make better-informed decisions.

Supporters counter that the ambiguity is the point: the law is designed to force companies to keep humans meaningfully accountable, whatever their particular tooling. Labor unions celebrated the signing as a national template, with the California Federation of Labor Unions hailing the first-in-the-nation guardrails as proof that human oversight and worker protections can be written directly into law.

Why It Matters Beyond California

For anyone building HR technology, management platforms, or “boss-ware” productivity trackers, California just set the most consequential compliance target in the U.S. The state’s economy rivals entire countries, and its technology regulations — from privacy to emissions — have historically become de facto national standards as multistate employers standardize on the strictest rule.

More broadly, SB 947 signals a shift in how AI regulation is being framed: not as a question of model capability or existential risk, but of accountability in everyday decisions. Illinois’ AI employment disclosure law, in effect since January, requires transparency but still allows AI to be the sole decision-maker. California has now drawn a harder line — one that other states, and eventually Congress, will be measured against.

The message from Sacramento is simple: AI can advise, analyze, and recommend. But when it comes to deciding whether a person keeps their job, a human must decide — and be able to say why.