California has made one principle clear: an algorithm may inform a workplace decision, but it cannot stand alone between an employee and a lost job. The new law’s promise of accountability, however, is matched by business concerns that its boundaries remain unclear.

Gov. Gavin Newsom’s measure bars employers from relying solely on automated decision-making systems to discipline or terminate workers. When AI is used as the primary basis for such an action, a human reviewer must corroborate the decision with additional information, including evaluations, peer reviews or personnel records. Workers must also receive written notice, a description of the data used and access to a human contact who can explain the outcome.¹

The conservative coverage frames the measure broadly as an effort to “protect workers from AI risks,” emphasizing its worker-safety rationale without endorsing the wider regulatory campaign.² The liberal account presents it more pointedly as the “No Robo Bosses Act,” portraying the law as a landmark victory for organized labor and a direct rejection of machine-only management.

Supporters argue that workplace software can reproduce bias or make consequential errors. “No worker should ever be fired or disciplined by a machine,” said Democratic state Sen. Jerry McNerney, adding that AI should remain “a tool controlled by humans, not the other way around.”¹ Labor leader Lorena Gonzalez similarly called the law evidence that workers and unions can “fight back and win” against AI’s expansion into the workplace.¹

Opponents’ concern is less about whether humans should remain involved than about when the law is triggered. Chamber of Progress representative Robert Singleton argued that “primarily relies” is undefined, potentially discouraging tools that improve consistency or identify safety risks.¹ Newsom himself vetoed an earlier version over broad notification rules, but revisions removed advance notice requirements and narrowed the bill’s reach. California has therefore chosen a compromise: strong scrutiny of high-stakes AI decisions, paired with a legal standard businesses say may still be difficult to interpret.