Governor Gavin Newsom surprised few with his veto of a bill targeting secret recordings made with smart glasses. Yet hours earlier he approved a measure that bars companies from letting algorithms fire workers on their own. The contrast captured California’s messy push to set boundaries on artificial intelligence in daily life.

On September 30, Newsom signed 13 AI-related bills. One stands out. SB 947, known as the No Robo Bosses Act, makes California the first state to prohibit employers from relying solely on automated systems for discipline or termination. A human must review the output. That person has to bring independent evidence such as manager evaluations, peer reviews or personnel files. Rubber-stamping the machine’s conclusion falls short. Workers receive notice when AI played a primary role, along with a description of the data used and a contact who can explain the decision. The rules take effect July 1, 2027.

“No worker should ever be fired or disciplined by a robo boss,” said state Sen. Jerry McNerney, the bill’s author, in a statement reported by The Next Web. He added that AI systems can boost productivity but remain prone to errors, bias and misjudgments. The legislation drew support from the California Federation of Labor Unions.

This outcome marks a reversal. Newsom vetoed a similar bill last year. He worried then that broad notification requirements would burden businesses using even routine software. The revised version narrows its focus to high-stakes decisions that affect livelihoods. It drops advance notice obligations and certain gig-worker protections that had drawn opposition from Uber and Lyft.

“I share the author’s concern that in certain cases unregulated use of ADS by employers can be harmful to workers,” Newsom wrote in his earlier veto message, according to CNBC. The governor has positioned himself as an AI watchdog in recent months. His actions come as public sentiment toward big tech has soured.

The package extends further. New laws limit AI-driven surveillance in bathrooms, bar predictions of workers’ emotional states, and restrict collection of neural data. They reinforce that health care professionals, not algorithms, hold final say over patient care. And they require disclosure when automation contributes to layoffs. Several measures Newsom rejected last year returned in tighter form and won approval this time, the San Francisco Chronicle reported.

Yet not every proposal survived. Newsom vetoed SB 1130, which aimed to penalize secret recordings made with wearable devices in private spaces such as changing rooms or doctors’ offices. The bill would have mandated a visible recording light on such devices starting in 2028 and banned sales of gear designed to hide that indicator. In his veto message he called the definition of “wearable recording device” too broad. It might sweep in ordinary smartwatches used in public, he said, creating confusion. Existing privacy statutes already cover non-consensual recording in private areas.

Tech industry groups including TechNet, whose members count Meta, Google and Amazon, opposed the smart-glasses measure. The veto drew fresh attention on social media as wearable cameras proliferate. Meta alone sold millions of its smart glasses last year. Similar products from Snap and smaller startups keep launching. Concerns about always-listening hardware have grown alongside them. A recent TechCrunch article detailed how the proposal sought to make California the first state to directly regulate the devices.

Newsom’s decisions arrived one day after he sharply criticized a White House meeting hosted by President Donald Trump. Tech executives gathered there and signed a voluntary AI safety accord. In a video posted to X, the governor dismissed the event. “What I heard in Washington DC yesterday is bullshit, and that should scare the hell out of everybody,” he said, per multiple reports including The Next Web.

The remark underscored a larger tension. Federal leadership on AI remains light. States, led by California, have filled the gap with targeted rules. Labor unions pressed hard for workplace protections. Business groups warned that overly prescriptive laws could slow innovation and raise compliance costs. The revised No Robo Bosses Act represents a compromise that both sides continue to debate.

Supporters argue the law closes a dangerous loophole. Automated decision systems now influence hiring, scheduling, performance scoring and more. Errors happen. Bias creeps in. Without human corroboration, workers risk losing jobs over flawed data or opaque logic. The notice requirement gives employees a chance to challenge outcomes and understand what information shaped them. But critics inside HR departments and technology vendors say the mandate could complicate tools already embedded in enterprise software. Real-time coaching features or predictive analytics might need redesign.

Newsom also turned down bills that would have shielded health care workers from retaliation when they override AI recommendations and restricted certain uses of AI in mental health settings. He called one proposal overly broad and questioned whether the Labor Commissioner possessed the medical expertise to enforce another. Those vetoes disappointed unions that viewed the measures as essential patient-safety guards.

California’s approach stands apart. Other states have passed narrower AI rules, often focused on deepfakes or transparency in government use. None match the breadth of workplace restrictions now on the books here. The new statutes build on existing privacy and labor frameworks rather than create an entirely separate regulatory structure. Enforcement will fall largely to the Labor Commissioner and courts.

Industry insiders watch closely. Human-resources software makers must now ensure their platforms support human-in-the-loop workflows that satisfy the corroboration test. Employers using third-party AI tools for performance management will review contracts and update policies. Larger firms already employ oversight teams; smaller ones may need new training. The July 2027 start date gives time to adapt. Yet questions linger about how strictly “primarily used” will be interpreted and what counts as sufficient independent evidence.

And the smart-glasses veto leaves a gap. Wearables keep advancing. Cameras grow smaller. Audio capture becomes always-on. Lawmakers may return with narrower language that targets only devices clearly designed for covert recording. Or they could strengthen existing consent and privacy statutes instead. The debate will not fade.

Newsom’s record this session shows evolution. He rejected fewer tech-regulation bills than in prior years. Public opinion, shifting against unchecked technology power, appears to have influenced the calculus. At the same time he avoided measures that risked broad economic disruption or unintended bans on useful tools. The result is a pragmatic mix of protections and restraint.

California workers now possess new rights when algorithms enter the termination process. They cannot be judged by machine alone. A person must look at the data, add context, and stand behind the call. That safeguard feels modest to some. To others it draws an essential line. Either way it sets a precedent other states may soon test. The era of pure robo bosses in the Golden State just ended. What replaces it will unfold in offices, courtrooms and legislative chambers in the years ahead.