California has passed a law banning employers from using artificial intelligence as the sole basis for firing or disciplining workers, imposing new human-review and notification obligations on any business with employees in the state.

The measure, SB 947 — known as the No Robo Bosses Act — blocks companies from exclusively relying on "automated decision-making systems" in disciplinary and termination matters, and restricts AI's supporting role in those decisions as well. Governor Gavin Newsom signed the act after state senator Jerry McNerney, the bill's author, first introduced it in 2025 following a union-backed campaign.

The law's compliance requirements are concrete. Employers must have a human reviewer corroborate AI-driven decisions using managerial evaluations, peer reviews and personnel files. Workers must receive written notice when AI was "primarily used" in a termination or disciplinary decision, along with a description of the employee data the system processed and a human point of contact who can explain the outcome.

For HR leaders, that translates into several people-process decisions: mapping which automated systems touch discipline and exit workflows, documenting the data those systems ingest, and designating accountable human reviewers. The undefined phrase "primarily used" is likely to be the compliance flashpoint — employers must judge for themselves when a tool crosses from informing a decision to driving it.

McNerney, a Democrat, framed the law's intent bluntly. "No worker should ever be fired or disciplined by a machine, AI or not," he told CNBC. "Artificial intelligence systems have the potential to boost productivity, but they've also made errors and misjudgments and exhibited bias."

Lorela — Lorena Gonzalez, president of the California Federation of Labor Unions, AFL-CIO and the act's lead sponsor, said after Newsom signed the bill: "When working people organise, we get results. Workers across California have demanded that our state lead the way in regulating AI in our workplaces. And today, we see that begin to happen."

The context: high adoption, rising litigation risk

The law lands amid unusually heavy workplace AI adoption. In the US, 90% of managers say their firms have adopted at least one tool to "instruct, monitor or evaluate workers" — a share thought to exceed that of businesses in any other country.

Litigation has already tested the territory. Meta faces a lawsuit, filed in July, in which former employees allege AI-assisted systems ranked and selected workers for layoffs, disproportionately affecting those who had taken medical or family leave. The workers claim the tools targeted them after they requested protected leave, maternity leave or disability accommodation. Meta has denied the allegations.

A retail sector survey in May found Walmart and Amazon workers registering growing concern that HR decisions were being automated.

A negotiated bill with carve-outs

Newsom initially objected to the act, arguing it was too unfocused and would hamper employers using AI in even innocuous settings. Lawmakers amended it to strip onerous requirements and shield rideshare platforms such as Uber and Lyft, which rely heavily on gig workers. Many California business leaders still opposed the measure in the weeks before enactment.

Robert Singleton, senior director of policy and public affairs for California and the US West at Chamber of Progress, a technology trade group, pressed the definitional problem in a letter to Newsom. "The bill's obligations generally apply when an employer 'primarily relies' on an automated decision system, but that critical term is never defined," he wrote. "Employers are given no objective standard for determining when a technology has moved from merely informing a decision to being a primary basis for it."

Public sentiment shifting against AI

The legislation tracks broader public unease. A July Gallup poll found 39% of Americans believe AI does more harm than good, up 8 percentage points from 31% in 2025. Pew Research, in a study released in August, found 71% of Americans think AI will take people's jobs — an increase of roughly 7 points from about two years ago.

Newsom has moved on adjacent fronts as well, signing an executive order this month on the potential existential risks of AI models and a measure creating a state framework for independent evaluation and auditing of AI systems. He accused the federal government of abdicating "its responsibility to protect Americans."

With SB 947, California becomes the testing ground for how far states will go in mandating human oversight of algorithmic management — and HR teams should expect the undefined "primarily used" standard to be worked out through enforcement actions and court challenges before employers get clearer guidance.