California Governor Gavin Newsom signed a package of laws on 30 September 2026 that puts human oversight at the centre of how employers may use AI on their workforces.

The measures require human review of AI-based decisions to fire or discipline an employee, rather than letting an algorithm's output stand as the decision. They add notice requirements when AI contributes to mass layoffs, and they place limits on workplace surveillance. The signing extends California's run of AI legislation: earlier this month Newsom signed a law requiring developers of large AI models to report safety incidents and disclose safety plans, and separate rules obliging companion chatbots to carry crisis protocols.

The workplace package lands in the middle of a live argument about automation and jobs. Companies are increasingly selling "workforce analytics" tools that score and rank employees, and regulators have worried that such systems can encode bias while giving managers a veneer of objectivity. Requiring a human in the loop does not ban the tools; it changes who is accountable when the tool's recommendation becomes the outcome.

Critics on both sides will find something to dislike. Employer groups have long argued that state-by-state rules create a patchwork of compliance. Labour advocates, meanwhile, may note that human review is only as good as the human doing it — a manager rubber-stamping an algorithmic ranking satisfies the letter of the requirement without changing the result. California is once again the test case: whatever enforcement looks like here tends to travel.

Sources