Governor Gavin Newsom signed SB 947, the No Robo Bosses Act of 2026, on September 30, making California the first state in the nation to prohibit employers from relying solely on automated decision-making systems to fire or discipline workers. The bill, authored by Senator Jerry McNerney, was filed with the Secretary of State the same day as Chapter 859 of the Statutes of 2026. The law takes effect July 1, 2027. The bill won approval in the Senate on a vote of 28-10 after the Assembly approved it 53-14. The path to enactment was not straightforward: Newsom reversed the veto he had issued in October 2025 on a near-identical version, which he had then called overly broad. McNerney brought the bill back in February and trimmed it, removing the pre-notification requirement that drew Newsom's objection and cutting language that would have covered gig workers, which had been opposed by Uber and Lyft.
The statute's core obligations are layered and precise. SB 947 bars California employers from relying solely on automated decision-making systems to fire or discipline workers, and mandates human oversight and verification when employers use such systems to assist in termination and disciplinary decisions. Employers that rely primarily on AI output must have a human reviewer corroborate the decision with additional inputs, which the bill lists as "managerial evaluations, peer reviews, and personnel files." Workers also gain procedural rights: anyone fired or disciplined under a system that primarily used AI must be given written notice that AI was in the loop, a description of the employee data the system drew on, and a human point of contact who can explain the decision. Workers may also request, and employers must provide, a copy of the most recent 12 months of the worker's own data primarily used by the system to make a disciplinary, termination, or deactivation decision. Enforcement falls to the Labor Commissioner and public prosecutors, with a $500 civil penalty per violation. Crucially, the law applies to public and private employers regardless of size, with no small-business exemption.
The statute's definition of what it regulates is deliberately broad. SB 947 defines an automated decision system as any computational process derived from machine learning, statistical modeling, data analytics, or artificial intelligence that issues simplified output, including a score, classification, or recommendation, which is used to assist or replace human discretionary decision-making and materially impacts natural persons. That language is wide enough to capture performance-management platforms, workforce-analytics dashboards, and attendance-flagging software that many employers do not think of as "AI." Newsom signed SB 947 alongside a package of other AI-related measures, including a bill requiring employers to notify workers in writing if a mass layoff is caused by an AI system.
Skeptics on both sides have raised substantive objections. Business groups argued during committee hearings that "the bill's obligations generally apply when an employer 'primarily relies' on an automated decision system, but that critical term is never defined," leaving compliance standards uncertain. The California Chamber of Commerce testified in opposition, contending that the corroboration language creates real compliance problems and that the data-access provisions are overbroad. Labor advocates and some legal analysts counter that the law's protections may be insufficient on their own: SB 947 is a real, meaningful first step that will likely function as a paperwork requirement rather than a substantive check until it is paired with transparency and protection provisions it currently lacks. Regardless, California's market size means national employers with operations in the state will likely adopt uniform policies rather than maintain two sets of HR rules, a dynamic that has played out before with state privacy and emissions standards. If SB 947 works as written, expect copycat bills in other states and renewed union demands that the same human-oversight standard be extended beyond firing and discipline into hiring, scheduling, and performance scoring.
The real test arrives July 1, 2027: whether a $500 penalty and a signature line on a review form are enough to stop a manager from treating the algorithm's output as the decision rather than merely a recommendation.