California just banned employers from firing workers on AI's word alone -- and told state agencies to keep saying "AI," not "Super Intelligence"

California Gov. Gavin Newsom signed a package of AI workplace bills on Sept. 30, the state's annual deadline for acting on legislation passed that session. The centerpiece, SB 947 -- titled the "No Robo Bosses Act" -- makes California the first state to bar employers from relying on AI as the sole basis for firing, disciplining, or deactivating a worker; a human must review and corroborate the decision first. A companion measure, SB 951, requires employers to give written notice when AI systems are responsible for mass layoffs or terminations. Two more bills, AB 1331 and AB 1883, restrict workplace surveillance tools -- including an explicit ban on using them in employee bathrooms.

Newsom signed the bills alongside the smaller, sharper-edged order: state agencies are to keep calling the technology "artificial intelligence," not the "Super Intelligence" terminology President Trump's Sept. 29 order directed the federal executive branch to adopt wherever legally possible. That federal order applies only to how the US government talks about AI -- not to what any company ships -- and Newsom's countermove is exactly as symbolic, with a sharper edge attached: "Super intelligence is clearly not coming from the White House," the state's announcement noted, pairing the jab with a worker-protection package the federal government has not matched with anything binding.

"AI should expand opportunity -- not come at the expense of workers and families." -- Gov. Gavin Newsom

The workplace bills are the headline, but they're one piece of a larger signing slate the same day. AB 1979 requires doctors using AI clinical-decision tools to retain their own professional judgment and requires developers to work on reducing algorithmic bias in those tools; a companion bill, SB 503, extends similar protections to AI used in companion-care healthcare settings. SB 574 bars attorneys from fully delegating core legal work -- brief-drafting among it -- to AI. AB 2713 and SB 1000 tighten the existing California AI Transparency Act's rules on disclosing a piece of content's AI provenance, and SB 1111 updates civil and criminal protections against unauthorized AI-generated "digital replicas" of a person's voice or likeness. None of these carries the same bright-line, first-in-the-nation claim SB 947 does, but together they're the clearest evidence yet that California is legislating AI's workplace, courtroom, clinical, and likeness-rights edges in parallel rather than waiting to pass one comprehensive law.

Newsom's own framing leaned on federal inaction: "we have to do a lot more in the absence of federal leadership," he said, leaving open the possibility of a special legislative session to go further. The sequence matters -- this is the third distinct California AI executive action since Sept. 18, after an order directing agencies to study a frontier-model kill switch and, earlier in September, a law requiring AI chatbot operators to run risk assessments before deployment. California isn't alone in moving at this pace, either: Illinois and Oregon signed their own AI executive orders a day apart in late September, though both bind only their own state governments' AI use, while SB 947 reaches into every private employer in California.

The reaction splits almost exactly along who the bills bind. "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," said Lorena Gonzalez, president of the California Labor Federation (AFL-CIO). The California Chamber of Commerce took the opposite read of the same bill, warning in its formal opposition letter that SB 947 "broadly targets businesses of all sizes, across every industry, and regulates even low-risk applications of automated decision systems," and that it "will drive up costs for consumers and employers because it would impose significant compliance burdens and any misstep would lead to costly litigation." The Chamber separately argued that SB 947's human-corroboration requirement "may be impossible or highly impractical" for employers to satisfy in practice -- a direct challenge to whether the law's central mechanism is even operable as written, not just whether it's wise policy. The Chamber of Progress, a separate tech-industry group, raised a narrower objection: that SB 947's key terms remain undefined.

What SB 947 does not do is ban AI from employment decisions -- it bans AI from being the *only* basis for one. An employer can still use an automated system to flag a performance issue or recommend a layoff; the law's actual mechanism is narrower than "No Robo Bosses" makes it sound, requiring a human to corroborate the system's output before acting on it, not requiring a human to make the decision independently. That distinction is exactly where the Chamber's practicality objection and the labor federation's victory framing both aim: whether "corroborate" becomes a meaningful check or a rubber stamp is a question the bill's text doesn't fully resolve, and one no court or regulator has yet tested.

Nothing in the Sept. 30 signing specifies when SB 947 takes effect, how it will be enforced, or what budget -- if any -- California is attaching to policing it; none of the sources for this piece supply those numbers, which is itself worth noting rather than guessing past. What's settled is the claim itself: California is now the only state where an employer cannot lawfully reduce a firing decision to an algorithm's output alone, and the next real test of that claim won't be a press release -- it will be the first worker who says a human's "corroboration" wasn't one.