California Governor Expands AI Safety Rules

California Governor Gavin Newsom has signed an executive order requiring independent verification of artificial intelligence models. The measure establishes a committee of experts focused on developing kill switches for AI agents that exhibit dangerous behaviors.

This action comes late in Newsom’s term and marks a shift in stance compared to previous proposals. In 2024, the governor vetoed a state bill that included similar safety requirements. At the time, he argued that such restrictions could limit the technology's development potential.

The new order seeks to expand the definition of security incidents and include measures that were previously rejected. The move occurs as Newsom shapes his political trajectory for a potential 2028 presidential bid. Analysts note that the initiative responds to growing concerns among American voters regarding the impacts of automation.

In October 2025, the state had already enacted a law requiring major developers to report serious incidents. Events requiring notification include large-scale cyberattacks, significant financial theft, or situations resulting in deaths. The current order reinforces oversight of the technology sector in the region.