California’s ‘No Robo Bosses’ Act Is Back — What It Means If You Screen Candidates for US Roles

California Governor Gavin Newsom vetoed the original “No Robo Bosses Act” in October 2025, calling it overbroad. Nine months later, it’s back. State Senator Jerry McNerney introduced SB 947, the No Robo Bosses Act of 2026, reviving a bill that would bar employers from letting an automated system make high-impact employment decisions, including hiring, without meaningful human review (Fisher Phillips, “California Set to Restrict AI Use in the Workplace”). If your team screens candidates for any US-facing role, this is worth tracking now, not after it either passes or dies a second time.

What actually changed between SB 7 and SB 947

The 2025 version, SB 7, would have required thirty days’ written notice before deploying an automated decision system, with disclosure of what data it uses and how it processes it (California Legislative Information, SB 7 bill text). Newsom’s veto message argued it duplicated existing regulation and was too broad for the range of tools it would have covered. SB 947 narrows the target: instead of every automated workplace tool, it goes after the specific case of AI making a high-impact call: firing, discipline, or hiring, without a human in the loop who can independently verify the outcome. That’s a deliberate response to the veto, not a repeat of it, and it means the bill has a real shot at surviving where SB 7 didn’t.

The pattern worth noticing

This isn’t an isolated California move. Illinois already requires disclosure whenever AI plays a role in an employment decision, and New York City’s Local Law 144 enforcement is heading into a stricter phase, not a looser one. California’s own FEHA regulations already cover automated decision systems. SB 947 is the next layer on top of that stack: even if a tool itself is compliant, the law increasingly wants a human who can override it. We’ve written before about what that documentation burden actually looks like when your legal team starts asking questions about an AI hiring tool. Human-oversight requirements are the next line item on that list.

Why India-based screening teams should care

A Global Capability Center or IT services firm running first-round AI voice screens for a California-headquartered client isn’t outside this conversation just because the recruiters sit in Bengaluru or Pune. The obligation attaches to the employment decision and the candidate the role belongs to, not to the physical location of whoever configured the screening campaign. If SB 947 becomes law, “the AI auto-rejected them and nobody looked at it” stops being a defensible answer for a California role, regardless of which country ran the screen.

The uncomfortable part for vendors selling full automation

A lot of AI screening pitches lean hard on “zero-touch” as the selling point: the promise that HR never has to look at a candidate until the shortlist is done. That pitch gets harder to make in a regulatory environment that increasingly assumes full automation on a high-impact decision is itself the risk, not just a biased outcome. The honest answer isn’t to abandon automation. Screening 60%+ of a candidate pool down to a clear no on obvious criteria is still the only way to make a 3,000-candidate evening tractable. It’s to be precise about which decisions the system makes outright and which ones it routes to a person, and to be able to show your work either way.

Newsom’s veto had a real point, too

It’s worth taking the 2025 veto seriously rather than treating it as noise. Newsom’s stated concern was that SB 7 was broad enough to sweep in routine scheduling and workflow tools alongside genuinely high-stakes automated decisions, which risked making compliance expensive without making anyone safer. That’s a legitimate design problem, not just political cover. A law that treats a shift-scheduling algorithm the same as an AI system deciding who gets fired doesn’t actually protect workers better, it just spreads paperwork thinner. SB 947’s narrower scope, limited to hiring, discipline, and termination, reads like an attempt to fix exactly that criticism rather than ignore it. Whether it’s narrow enough to pass is still an open question, but it’s a more defensible bill than the one Newsom rejected.

What to check before your next US-facing drive

If you’re screening for Illinois, California, or New York roles from India, three things are worth confirming this quarter: whether your vendor can tell you exactly which decisions the AI makes unsupervised versus which ones a recruiter reviews, whether that split is documented anywhere you could hand to a client’s legal team, and whether “human oversight” in your current process means a real review or a rubber stamp on a pre-decided outcome. SB 947 hasn’t passed yet. The direction of travel has been consistent for two years running, and betting against it has gotten more expensive every cycle. If you want to see how we split automated decisions from human review on a live campaign, book a demo.