Gavin Newsom signed an executive order to require AI labs like Open AI and Anthropic to implement a "kill switch", signaling growing political concern. The order sets up experts to work out AI safety measures, including legally requiring shutdown ability. Newsom, a likely 2028 Democratic presidential candidate, is toughening his stance ahead of midterm elections. "While Washington abdicates its responsibility to protect Americans, California is building on the strongest AI regulatory framework in the nation," he said.

Calls for a "kill switch" follow warnings from leading US AI labs to slow development. Experts debate feasibility and whether advanced AI could overcome such measures. President Donald Trump rejected slowing AI, arguing the US must beat China and safety concerns stem from "negative forces". He has tried to block state-level AI regulations. Other Republicans, including House Speaker Mike Johnson, urged calm and echoed Trump on China. Federal "kill switch" laws are unlikely to advance before November's elections.

The concern follows Anthropic researcher Jacob Coxon's resignation, warning AI "could kill us all by the end of the decade". Leaders Dario Amodei of Anthropic, Sam Altman of Open AI, and Space X's Elon Musk made a rare joint call to slow rapid model capability advances after hacking incidents. Amodei proposed embedding independent evaluators. Newsom said California would examine making this a legal requirement and having third-party safety plans. He also ordered faster implementation of two California AI safety laws. Open AI welcomed the move, pledging to work with the governor to "advance strong, adaptable safeguards".