California just drew a hard line that most states have not even attempted: a computer cannot be the final word on whether you keep your job. Governor Gavin Newsom signed SB 947, the "No Robo Bosses Act," banning employers from relying solely on AI to fire or discipline a worker. A human has to confirm the decision, backed by real evidence like a manager's evaluation or a personnel file.
This is not Newsom's first attempt. He vetoed a near identical bill, SB 7, in October 2025 after business and tech lobbying groups pushed back hard. The revived version passed anyway this time, with labor unions applying pressure through the legislative session.
The law does not take effect until July 1, 2027. That gives companies time to build the paperwork trail they will need: documented human sign off, corroborating records, and written notice to any worker affected.
A companion law updates California's long standing mass layoff notice rule, the Cal/WARN Act. If a company lays off a large group of workers and AI or automation is a major reason, it now has to say so in the notice. That detail matters more than it looks.
Once a company admits in writing that AI drove a layoff, that document becomes evidence in future wrongful termination claims, union negotiations, or public pressure campaigns. Expect employers to get very careful about how they word these notices.
A third rule bans using biometric data, things like facial expressions, voice tone, or heart rate, to guess a worker's emotional state. This closes off a category of workplace monitoring software that has been quietly growing for years, often marketed as a way to spot burnout or flag safety risks.
The more interesting story is the fight happening above all of this. On the same day, Newsom signed an executive order telling state agencies to keep using the term "artificial intelligence" instead of "super intelligence," a term the Trump administration recently pushed U.S. diplomats to adopt. It is a small symbolic jab, but it points to something bigger.
The federal government has been moving toward encouraging AI companies to regulate themselves, including a voluntary pact tech CEOs signed with Trump just a day earlier. California is moving in the opposite direction, writing specific legal bans into state law.
This split is not going to resolve quietly. Reports have already surfaced of a federal push to get the Department of Justice to challenge state AI laws it considers too burdensome, with a federal framework floated earlier this year recommending that federal rules override state ones. If that fight reaches the courts, California's new worker protections could become the test case.
For now, if your company operates in California and uses any software to help with hiring, performance reviews, scheduling, or termination decisions, the clock is already running. Waiting until 2027 to figure out your compliance paperwork is a bad plan. Start mapping which decisions touch AI and which ones have a documented human check right now, because retrofitting that process under deadline pressure is far more expensive than building it early.