California Bans Firing by Algorithm Alone as Newsom Signs Workplace AI Package

California Governor Gavin Newsom signed a package of workplace AI bills on 30 September, the last day of his signing window. The centrepiece, SB 947 from state senator Jerry McNerney and billed as the No Robo Bosses Act, bars employers from relying solely on automated decision systems to discipline or fire workers and requires a human to corroborate such decisions using other information. SB 951 amends the state's Cal/WARN layoff-notice law so employers must disclose when job losses are caused in substantial part by AI. AB 1883 bans tools that infer workers' emotional states or collect neural data. Newsom framed the package as making sure AI does not come at the expense of workers and families.

Second time lucky

This is a do-over. Newsom vetoed a broader version, SB 7, in October 2025, objecting to notification requirements he considered overly broad. The new package is narrower and more targeted, and that is probably why it survived. Rather than regulating every automated hiring and management tool, it carves out three specific behaviours: firing by algorithm alone, hiding AI as the cause of layoffs, and reading employees' emotions from biometric or neural signals.

Business groups opposed the bills through the legislature. The California Farm Bureau, for one, opposed all three, arguing that AB 1883 in particular would limit common workplace safety, security and management tools. Those objections did not change the outcome, but they will shape the litigation. Phrases such as "solely" and "in substantial part" will be fought over in court long before anyone agrees what they mean in practice. SB 951 and AB 1883 take effect on 1 January 2027; SB 947 follows on 1 July 2027, which gives vendors nine months to adapt.

What it means for people who build HR software

For vendors, the requirements are concrete. A performance-management product that auto-generates termination recommendations now needs a workflow where a human reviews other evidence, and a record that this happened. An "attrition prediction" feature that feeds a reduction in force may trigger disclosure duties for the customer, which in turn means customers will start asking vendors to document what their models did. Sentiment and "engagement" analytics built on webcam, voice or wearable signals become hard to sell in California at all.

None of this is exotic engineering. Audit logs, human-in-the-loop steps and clear model documentation are things good teams should already have. The difference is that they now carry legal weight in the world's fourth-largest economy, and vendors rarely want to maintain a separate California version of their product.

The federal collision

The package lands in direct tension with Washington. A December 2025 executive order set up a Justice Department task force to challenge state AI laws and tied some federal broadband funds to states not enforcing what it calls onerous rules. Congress has so far declined to pass a moratorium on state AI legislation; the Senate stripped one out 99 to 1 in 2025. Newsom used the signing to criticise the absence of federal rules, and alongside the bills he issued an executive order telling state agencies to say "artificial intelligence" rather than "super intelligence".

Worth watching: whether the federal task force picks California's employment laws as a test case. Employment is traditionally state territory, which makes it a weaker target than, say, rules on model development. If Washington does not challenge these, expect other states to copy them quickly.

---

Sources

• California Gov. Gavin Newsom signs laws to protect workers from AI risks (AP via AJC)
• Newsom signs slate of AI workplace laws, barring 'robo bosses' and surveillance (KQED)
• Newsom signs revamped 'No Robo Bosses Act' into law (SHRM)
• California amends its WARN Act again (Littler)
• Trump executive order seeks to rein in state AI laws (GRC Report)