California bans AI-only firings as Newsom signs No Robo Bosses workplace laws

California Governor Gavin Newsom signed a package of first-in-the-nation AI workplace laws on Wednesday, including SB 947 that prohibits employers from relying solely on automated decision systems to discipline or fire workers.

Companies are already using software to score, schedule, and sometimes cut workers. California just wrote a hard line: a machine alone cannot fire you, a human has to own the call, and an AI-driven mass layoff has to be labeled as one, with the human-review rule in the signed bill starting July 1, 2027. That is the clearest state pushback yet against “the algorithm decided,” in the same week Washington pushed voluntary self-policing.

On Wednesday, 30 September 2026, Governor Gavin Newsom signed a package of laws the Office of the Governor calls first-in-the-nation protections for California workers. The announcement is on the governor’s site. It prints Sep 30, 2026. It does not print an hour. The dateline is Sacramento. The office says the new laws protect people at work from artificial intelligence, automated technology, and AI-powered workplace surveillance. The same page says he also signed an executive order permanently declaring that artificial intelligence is to be called “Artificial Intelligence” in California. The subhead calls that order “declaring the obvious.” Newsom is quoted on the page: “Super intelligence is clearly not coming from the White House – that’s why California continues to lead. AI should expand opportunity – not come at the expense of workers and families. As this technology reshapes the workplace, California is putting people at the center, ensuring we all have a voice in the decisions shaping their future.” That quotation is his, in the announcement.

The worker section leads with SB 947. The signed list titles it “Employment: automated decision systems,” by Senator Jerry McNerney, a Democrat from Pleasanton. The office’s summary says the bill ensures real people review automated employment actions by prohibiting employers from only relying on AI when making a disciplinary action or a termination decision. A disciplinary action is a formal punishment at work, short of firing. A termination is a firing. Those lines are the Governor’s office. The enrolled bill text, which the office says he signed, is where the rule is written out.

The enrolled text of SB 947 was published 4 September 2026. Legislative Counsel’s digest says that beginning on July 1, 2027, an employer may not use an automated decision system for certain functions, and that the purposes and the manner of use are limited. Section 1526.7 says this new part of the Labor Code becomes operative on July 1, 2027. Operative means the rules start to apply on that date. The signature on 30 September does not, on this text, make them apply the next morning. An automated decision system, which the bill shortens to ADS, is a computational process drawn from machine learning, statistical modeling, data analytics, or artificial intelligence. It issues a simplified output, such as a score, a classification, or a recommendation, used to assist or replace a person’s own judgment, when that output materially affects a person. The bill says the definition does not include a spam filter, a firewall, antivirus software, an identity and access tool, a calculator, a database, or another compilation of data. Those lines are the enrolled bill.

The firing rule is in section 1522. An employer shall not rely solely on an ADS when making a disciplinary or termination decision. Solely means the software’s output is the only basis. If an employer primarily relies on an ADS output for that decision, the employer shall direct a human to corroborate it. Corroborate means check it against other information. The bill’s examples are supervisory or managerial evaluations, personnel files, the employee’s work product, peer reviews, and witness interviews, which may include relevant online customer reviews. If the employer cannot corroborate the output, or the human reviewer concludes that the output is inaccurate, incomplete, or misleading, the employer shall not use that output to discipline or fire. The enrolled text uses the phrase “primarily relies.” It does not print a percentage, or any other numeric test, for when reliance becomes primary. Those lines are the bill.

When an employer did primarily rely on an ADS, section 1524 requires a written notice at the moment the employer tells the employee about the decision. The notice has to be plain language, a separate communication, in the language the employer already uses for routine messages, and easy to receive, including by email or a link. It has to say that the employer primarily relied on an ADS, that a human reviewed the decision and corroborated the output, how to reach a human for more information, and that the employer may not retaliate if the employee uses the rights in this part. The employee may also ask for a meaningful, objective description of their own data that the system used. That description has to hide other people’s personal information. The Labor Commissioner may enforce the part, including by issuing a citation. A public prosecutor may bring a civil action. An employer who violates the part is subject to a civil penalty of $500 per violation. Five hundred dollars is the amount printed per violation. It is not a cap on a lawsuit’s other remedies. The bill’s definition of employer includes a private employer and also state and local government, the California State University, community colleges, school districts, and the University of California if the regents agree, plus labor contractors. Those lines are the bill. They are not a ban on every scheduling tool or scoring tool.

Two limits sit in the same text. A union contract can set this part aside if the waiver is explicit and clear, the contract covers wages, working conditions, and other terms of work, and it provides protection from algorithmic management. Algorithmic management, in that line, is the bill’s phrase for software that directs or judges the work. Separately, the part does not apply to an automated decision system to the extent a federal statute, a federal regulation, or a binding federal contract requires that use for developing aircraft for the national airspace, or for products or services for national security, military, space, or defense, and only for those covered operations. That limit is written into the California bill. It is not a federal statute that cancels the rest of SB 947. Those lines are the enrolled text.

The same announcement names other worker measures signed that day. SB 951, by Senator Eloise Gómez Reyes, a Democrat from Colton, is listed as “Employment: technological displacement: notice.” The office summarizes it as more transparency, requiring employers to provide if a mass layoff, relocation, or termination is caused by an AI system. The bullet does not add a form, a headcount, or a number of days. AB 1331, by Assemblymember Sade Elhawary, a Democrat from South Los Angeles, is listed as workplace surveillance. AB 1883, by Assemblymember Isaac Bryan, a Democrat from Los Angeles, is listed as workplace surveillance tools. The office says the signed measures protect employee privacy by banning the use of workplace surveillance tools by employers in a bathroom located in the workplace. The page does not say which of the two bill numbers carries the bathroom rule. Lorena Gonzalez, president of the California Federation of Labor Unions, AFL-CIO, is quoted speaking on SB 947, SB 951, AB 1331, and AB 1883. She said workers had demanded that the state lead on regulating AI at work, and that by signing these four bills the governor had begun to put guardrails on AI at work: human oversight, limits on surveillance, and workers first. The page also prints her closing line as “We must continue to push for regulations to ensure we control this no technology, not be controlled by it.” Those lines are the Governor’s office.

The same signing list goes past the workplace. AB 1979, by Assemblymember Mia Bonta, a Democrat from Oakland, and SB 503, by Senator Akilah Weber Pierson, a Democrat from San Diego, are both titled “Health care services: artificial intelligence.” The office says doctors and licensed providers can use their own professional judgment when AI or other clinical decision tools are used in patient care, and that developers of those tools must take reasonable steps to reduce known or predictable bias in how the systems are used. SB 574, by Senator Thomas Umberg, a Democrat from Santa Ana, is the attorney bill. The office says it keeps lawyers responsible for practicing law by prohibiting them from fully handing over core legal work, such as drafting briefs or providing legal judgment, to AI. On transparency, the office says the new protections prohibit deleting a digital watermark and allow access to metadata, so it is easier to tell whether material was made by AI or by a person. A watermark, here, is a mark in a file that says how it was made. Metadata is the extra information attached to that file. The office also says it is strengthening the California AI Transparency Act, and that public comments at local government meetings must come from real people, not AI bots. The signed list names AB 2713, by Assemblymember Buffy Wicks, a Democrat from Oakland, as “California AI Transparency Act: system provenance data,” SB 1000, by Senator Josh Becker, a Democrat from Menlo Park, as “California AI Transparency Act,” and SB 1159, by Senator Christopher Cabaldon, a Democrat from Yolo, as “Artificial intelligence: transparency and governance.” Provenance is a record of where something came from. The page groups those transparency rules and lists those titles. It does not print a line that assigns each bullet to one number. AB 1864, by Assemblymember Marc Berman, a Democrat from Menlo Park, covers gene synthesis equipment manufacturers and providers. Gene synthesis is the making of genetic material to order. The office says those companies have to follow safety guidelines, verify who their customers are, and check what genetic material they send out, including for research on diseases such as polio and Ebola. The same list also includes AB 2392, on AI training and procurement at public colleges, and SB 1111, on digital replicas. Those lines are the Governor’s office.

CNBC and Bloomberg Law reported the signing the same evening. Paxton Honerkamp at CNBC, published Wednesday, 30 September 2026, at 7:29 p.m. Eastern, used the headline “California Gov. Gavin Newsom bans AI ‘robo bosses’ in landmark state law.” CNBC calls SB 947 the No Robo Bosses Act. It says the law stops employers from exclusively using automated decision-making systems to fire or discipline workers, and restricts the use of AI as a principal tool in those decisions. It says that when an employer relies primarily on an AI output, a human reviewer has to corroborate the decision with other information, such as managerial evaluations, peer reviews, and personnel files, and that the affected employee gets written notice that AI was primarily used, a description of the employee data the system used, and a human to contact. Senator McNerney told CNBC: “No worker should ever be fired or disciplined by a machine, AI or not.” He said the systems can boost productivity and have also made errors, misjudgments, and shown bias, and that “AI must remain a tool controlled by humans, not the other way around.” CNBC also reports that Newsom vetoed an earlier version last October, after it had passed both houses, and that the bill he signed dropped a requirement to warn workers in advance whenever an AI system could affect working conditions, and dropped language that would have extended the protections to gig workers. Those sentences are CNBC’s account of the rewrite. Chris Marr at Bloomberg Law, published 30 September 2026 at 11:17 p.m. UTC, writes that SB 947 bans businesses from making employee discipline decisions solely based on artificial intelligence or algorithmic tools, effective July 1, 2027, and that employers must give affected workers notices about the AI tools used and summaries of personal data considered. Bloomberg Law calls the measure a revised version of the No Robo Bosses Act that Newsom vetoed in 2025. The July 1, 2027 date in that sentence is Bloomberg Law’s, and it matches section 1526.7 of the enrolled bill. Neither report says the rules apply across California on 1 October 2026.

The picture is the Office of the Governor’s announcement graphic for this release. A dark blue field. Gold type at the top reads “CALIFORNIA’S NATION-LEADING.” Large white type under it reads “AI FRAMEWORK,” then “GOT STRONGER.” A gold line. Then: “GOVERNOR NEWSOM SIGNS MORE FIRST-IN-THE-NATION WORKER PROTECTIONS AND MORE.” The state seal is in the top right corner. It is the official graphic. It does not print a calendar date.

In plain terms, on Wednesday the governor signed SB 947. His office describes it as a bar on relying only on AI to discipline or fire a worker. The enrolled bill says an employer may not rely solely on an automated decision system for that decision. If the system’s output is the primary basis, a person has to check it against other information, or the output cannot be used. The worker gets a written notice when the employer primarily relied on the system. The same text says those rules become operative on July 1, 2027. The same announcement says employers have to say if a mass layoff, relocation, or termination is caused by an AI system, and that workplace surveillance tools are banned in a bathroom at work. The package also covers clinical judgment, a lawyer’s core work, AI watermarks, customer checks for gene synthesis, and an order that California keep the name Artificial Intelligence. It is California law. It is not a nationwide ban, and it does not make AI hiring or scheduling tools illegal the morning after the signature.