California Enacts 'No Robo Bosses Act' as First State to Require Human Sign-Off on AI Firings
California employers can no longer let an algorithm make the final call on firing or disciplining a worker. Governor Gavin Newsom signed SB 947, the No Robo Bosses Act, on September 30, 2026, making California the first U.S. state to require human review when automated systems drive discipline or termination decisions. The bill headlines a wider package of AI measures covering mass-layoff notices, workplace surveillance, legal practice, deepfake protections and the state's formal definition of artificial intelligence.
The signing reverses Newsom's earlier veto of the same proposal, a turnaround that gives the No Robo Bosses Act unusual weight. Employers in the state now operate under rules their governor had rejected months earlier. SB 947 was carried by Senator Jerry McNerney, and its core requirement is narrow but binding: a company may not lean solely on an automated decision-making system to discipline or dismiss someone, and a person must review the decision before it takes effect.
The law does not forbid companies from using AI to flag performance problems, rank employees or recommend action. It forbids letting the software finish the job alone. The text also restricts automated systems from acting as the principal tool in discipline and termination, which pushes employers toward documented human judgment rather than a rubber-stamp approval.
The veto reversal is the sharpest signal in the package. Newsom had blocked the robo-bosses measure once; the version he signed pairs it with limits on AI in law practice, media likeness and biology. Employers that read the earlier veto as a reprieve now have a law with a broader footprint than the bill it replaced.
What the No Robo Bosses Act Changes in Practice
HR platforms and enterprise software vendors that sell algorithmic management tools sit closest to the new rule. A product can still surface a ranked list of termination candidates; what changes is the evidence trail. Customers must be able to show that a person weighed the facts and made the call, which turns audit logging, escalation workflows and human-in-the-loop review from optional features into procurement requirements.
That shift carries cost. Compliance teams will need records showing who reviewed what, when, and on which inputs. Vendors that cannot produce those records risk losing enterprise deals, and buyers gain a concrete reason to demand human-review features written into contracts rather than accepted as add-ons.
Vendor differentiation will move accordingly. Human-review tooling, audit logging and explainability features become sales arguments in California accounts, while platforms built purely around automated scoring lose their most aggressive pitch. The competitive gap will show up first in contracts with large employers, where legal teams inspect AI features before deployment rather than after.
SB 951 handles the other end of the employment relationship. The measure amends California's Worker Adjustment and Retraining Notification requirements so that employers must give notice when a mass layoff, relocation or termination is caused in whole or substantial part by an AI system. The wording matters: a layoff framed as a restructuring still triggers disclosure if software drove the decision.
The layoff-notice change carries a reporting dimension as well. Employers must state when an AI system is a cause of a mass layoff, which gives state labor authorities a running record of where automation replaces workers. No such record existed before, and it will shape how regulators and researchers measure AI's effect on employment.
Two further limits target monitoring rather than decision-making. The package bans workplace surveillance tools in employee bathrooms and restricts the use of AI to predict a worker's emotional state from biometric data. Emotion-recognition software has been pitched for productivity and safety monitoring, and California now treats biometric inference about how a worker feels as a restricted category.
For monitoring vendors, those two provisions remove product categories outright in California. Neither depends on whether a human reviews the output; the restriction applies to the collection and the inference itself. That makes them harder to engineer around than the discipline rule, which employers can satisfy with process and paperwork.
Lawyers, Deepfakes and Gene Synthesis
Four additional measures round out the signing. AB 1331 requires legal professionals to retain responsibility for core legal work even when generative AI handles research, drafting or court filings. AB 1883 establishes safety guidelines for gene-synthesis providers. AB 2713 updates likeness and deepfake protections. SB 1000 formally defines artificial intelligence in California statute, a step that matters because older enforcement language assumed the term needed no definition.
| Measure | Subject | Core requirement |
|---|---|---|
| SB 947 | Workplace discipline | Human review required; no sole reliance on automated systems |
| SB 951 | Mass layoffs | Notice when AI causes a layoff, relocation or termination in whole or substantial part |
| AB 1331 | Legal practice | Licensed professionals retain responsibility for core legal work |
| AB 1883 | Gene synthesis | Safety guidelines for providers |
| AB 2713 | Deepfakes | Updated likeness protections |
| SB 1000 | Definition | Formal statutory definition of artificial intelligence |
The legal-practice provision may be the fastest to bite. AB 1331 places the duty on the licensed professional rather than the tool, so a firm that deploys AI drafting assistants must still assign a qualified human to own the work product. For law firms, that is a supervision requirement layered on top of existing professional-conduct rules, not a ban on the software.
AB 2713 extends likeness protections at a time when synthetic media tools can clone a face and a voice from a handful of samples. The measure updates California's existing rules instead of creating a new enforcement track, which keeps it consistent with the state's habit of amending statutes rather than writing fresh ones. AB 1883 applies the same logic to gene synthesis, where providers receive safety guidelines rather than a prohibition.
SB 1000's definitional clause is the quiet piece. Writing a statutory definition of artificial intelligence settles scope questions that enforcement agencies, courts and regulated industries would otherwise litigate case by case. Every other measure in the package depends on knowing what counts as an AI system, which makes the definition the load-bearing element even though it changes no conduct on its own.
What This Means for Employers and Vendors
California is now the only U.S. state requiring human review of AI-driven discipline and dismissal. Companies with operations in the state face a two-part burden: keep the automation, and prove a person signed off. Firms selling algorithmic management tools face a matching demand, since their customers will need reviewable records to demonstrate compliance.
For buyers, the calculation is straightforward. A workforce tool that cannot log human review and produce an audit trail is a liability in California, which gives procurement teams leverage to renegotiate contracts and timelines. Multistate employers should also expect the requirement to migrate. California's framework has repeatedly set the template other legislatures copy, and the layoff-notice rule in SB 951 is the kind of provision that spreads quickly because it reuses existing WARN machinery rather than building new enforcement bodies.
The package stops short of banning AI at work. What it does is keep a human in the loop and require workers to be told when software shapes decisions about their jobs. That distinction matters for how employers frame their compliance programs: the goal is documented oversight, not withdrawal from automated tools.
Timing adds pressure. The signing landed on the last day of September, giving compliance and legal teams a single package to absorb rather than a series of separate deadlines. Employers that already document human review for other employment decisions will find the adjustment smaller than those running fully automated pipelines.
Why this matters
California has drawn a line most employers had not planned around: automation can inform a firing, but it cannot deliver one. For companies running algorithmic management at scale, the near-term work is documentation, not dismantling. For vendors selling those systems, human review is now a product requirement in the largest state market. The package's breadth, covering workplace rules, legal practice, deepfakes and gene synthesis, points to sector-by-sector regulation rather than a single AI statute.
Related Articles
- California's No Robo Bosses Act would end AI-only firings as 6 in 10 managers admit using AI in layoffs
- California AI Audit Laws Signed as Newsom Pushes for National Rules
- California AI Kill Switch Order Fast-Tracks Onsite Audits of Frontier Labs
✔Human Verified
Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.