California sets a November deadline for AI safety proposals

California Governor Gavin Newsom signed Executive Order N-9-26 on September 18. The order gives state agencies until November 16, 2026, to propose changes to AI safety law.

The agencies must consider letting outside evaluators check AI companies' safety reports for themselves. They must also consider a kill switch that outside reviewers would test over time. The review covers reporting rules for incidents in which developers lose control of their systems.

The order asks agencies to consider kill-switch requirements. It does not require every AI developer to install one now. The review must still decide whether the technology is feasible.

After the November deadline, lawmakers will have proposals to consider. The order does not yet say what a kill switch must stop or how an outside reviewer would test it.

Companies buying AI services need proof that the shutdown procedure works. They also need to know who is responsible when it fails. A product description that promises human control does not answer either question.

The recommendations are due November 16. By May 1, 2027, California must also publish criteria for organizations that want to conduct independent verification.

Watch whether California defines a shutoff that reviewers can test and lets evaluators report failures. Those decisions will determine how much power outside reviewers have.