California bars employers from firing or disciplining workers on an AI system’s output alone
A new California law requires a person to check an automated system’s conclusion before it leads to discipline or dismissal, and workers to be told it was used. It takes effect July 1, 2027.
On September 30, 2026, California Governor Gavin Newsom signed a package of AI bills, several of them about work. The central one, SB 947, stops employers from relying only on an automated decision system to discipline or fire someone. A person has to corroborate the system’s output by reviewing the data it used and other relevant records, such as performance evaluations and personnel files. If the output can’t be verified that way, the tool can’t be used for the decision. Workers must be told when such a system played a part, and they can ask for a description of their own data the tool used. KQED reports that the law takes effect on July 1, 2027.
Other bills in the package require employers to give notice of mass layoffs, relocations or dismissals caused by AI, and restrict workplace surveillance tools, including a ban on using them in bathrooms at work. Newsom vetoed a broader bill on the same subject, SB 7, in 2025. On September 30, he also vetoed a bill under which public employees would have been told when AI takes on tasks that belong to their jobs.
What it means for your work
California has now written into law a safeguard that HR teams and managers anywhere can apply when AI helps with performance reviews or discipline: a person checks the system’s conclusion before it affects someone’s job.
In practice, keep a record of who reviewed the system’s output and what evidence they looked at, and tell the employee that a tool was involved. A reviewer who simply approves what the system suggests has not corroborated anything. Employers with staff in California have until July 1, 2027, to put this in place.
Sources
This page is educational and is not legal advice.