California Governor Gavin Newsom has signed the “No Robo Bosses Act” into law, establishing a first-in-the-nation prohibition against employers relying exclusively on artificial intelligence to discipline or terminate staff. The legislation, designated as SB 947, represents a significant regulatory shift in workplace AI oversight and marks a reversal of Newsom’s stance after he vetoed an earlier version of the bill last year.
Under the new statute, businesses within the state are barred from using automated decision-making systems as the primary factor in employment adverse actions. If an employer primarily relies on AI output for termination or disciplinary decisions, they must now engage a human reviewer who corroborates the decision using supplementary data, such as personnel files, peer reviews, and managerial evaluations. Additionally, affected workers are entitled to written notice detailing the use of AI, a description of the data utilized by the system, and a designated human contact for further explanation.
Senator Jerry McNerney, the Democrat who authored the bill, emphasized that AI should serve as a controlled tool rather than an autonomous decision-maker. “No worker should ever be fired or disciplined by a machine, AI or not,” McNerney told CNBC. “Artificial intelligence systems have the potential to boost productivity, but they’ve also made errors and misjudgments and exhibited bias. AI must remain a tool controlled by humans, not the other way around.”
The law’s enactment follows intense lobbying by organized labor and comes amid growing public skepticism toward automated management. A July Gallup poll indicated that 39% of Americans believe AI does more harm than good, while Pew Research found that 71% of Americans fear AI will lead to job losses. These sentiments have intensified as Newsom considers a 2028 presidential run, with AI policy expected to become a central campaign issue.
Newsom initially vetoed the bill in October after it passed both legislative chambers, citing concerns over notification requirements that would have compelled businesses to warn workers in advance whenever AI systems were in use. In response to his objections, McNerney reintroduced the legislation in February, removing the pre-notification mandate and stripping provisions that would have extended protections to gig workers—a change that addressed criticism from rideshare companies like Uber and Lyft.
Despite these modifications, business groups continued to oppose the measure. Robert Singleton, senior director of policy and public affairs for the Chamber of Progress, urged Newsom to veto SB 947, arguing that the term “primarily relies” lacks a clear definition. He warned that uncertainty about which technologies qualify as regulated automated systems could discourage employers from adopting tools that enhance safety and consistency.
Labor leaders celebrated the signing as a victory for workers’ rights. Lorena Gonzalez, president of the California Federation of Labor Unions, stated, “When working people organize, we get results. Workers across California have demanded that our state lead the way in regulating AI in our workplaces.” Notably, Steve Hilton, the Republican nominee for governor, also voiced support for the act, arguing that current regulations do not go far enough to prevent AI from determining employment outcomes.
The passage of SB 947 positions California as the most prominent jurisdiction with strict workplace AI oversight. While Illinois enacted a law requiring employer notification for AI use in employment decisions earlier this year, it falls short of California’s outright ban on sole reliance on automated systems. At the federal level, similar legislation introduced by Senators Ed Markey and Brian Schatz has faced stagnation in Congress. However, California’s action may serve as a catalyst for other states, including New York, Louisiana, and New Jersey, which have pending proposals to regulate algorithmic management.
But what defines ‘primarily relies’? The vague wording worries me. Employers might just hide behind humans to keep using biased AI.
This is exactly the guardrail we needed. Letting algorithms单方面 decide a human’s livelihood is just dangerous and outdated.