Government Off Button? California Moves

Gavin Newsom speaks at a press conference with microphones outside a building — Photo: Jana Asenbrennerova / Shutterstock

California’s governor moved to design a government-verified “AI kill switch” for powerful models, reviving a controversial idea he rejected two years ago.

Story Snapshot

Gov. Gavin Newsom ordered work on an emergency “kill switch” for frontier AI models, with independent verification of its effectiveness.
The directive forms a panel to propose stronger state AI safety rules within two months, including third-party evaluators.
Newsom vetoed a 2024 bill that sought similar shutdown controls, after major tech groups warned it would hurt innovation.
Supporters frame the switch as part of a larger safety system; critics warn of vague rules and government overreach.

Newsom’s New Order Centers a Government-Checked AI “Off Switch”

Governor Gavin Newsom signed an executive order to “advance the creation of a kill switch” for advanced artificial intelligence models. The order says an independent verification group must test and check the switch on an ongoing basis. The plan places the shutdown tool inside a wider safety program, which includes incident reporting and stronger oversight. The governor’s office cast the move as a push to get ahead of fast-moving risks tied to powerful systems.

The order directs agencies to deliver a policy guide in about two months. That guide could recommend rules that require external safety evaluators at companies building top-tier models. The action aims to speed up ongoing state efforts and tighten gaps in company-run safety plans. This structure reflects a growing view among regulators that emergency controls may be needed if a model behaves outside human control or creates major security risks.

The 2024 Veto Still Hangs Over the Debate

In 2024, Newsom vetoed Senate Bill 1047, which sought to hold large model developers to strong safety standards and required a form of “full shutdown.” He argued the bill could push firms out of California and slow innovation. Major companies and tech groups warned the rules would harm the state’s economy and weaken national security, and they urged a veto. The move drew sharp criticism from safety advocates who said the biggest firms should face real accountability.

That history matters now. The new order echoes parts of the old bill’s goals but routes them through executive action and a short review window. This time, the state ties the shutdown idea to independent verification and broader incident response. The approach may avoid some details that helped sink the bill, while still giving the state leverage to shape the largest systems. The risk for Californians remains simple: rules that are too weak fail, and rules that are too vague invite abuse.

Security Promise Or Government Overreach? The Stakes For Freedom And Innovation

Backers say a kill switch could stop a dangerous model fast, when human control is lost or critical systems are at risk. They argue the switch must be tested by outside experts and checked over time, so it works when needed. That claim mirrors standard safety playbooks for high-risk tech. Supporters frame this as common sense: plan for failure, drill the response, and document who can press the button and when.

Critics warn a broad shutdown tool in government hands can be misused. Vague triggers and secret reviews can chill speech, crush small firms, and centralize power. Industry groups say heavy rules will drive jobs out of California and hand wins to rivals abroad. Conservative readers should track how the state defines “loss of control,” who decides when to shut down, and how due process works. Clear limits, public standards, and tight oversight must come first, not after the switch is built.