Newsom Executive Order Fast-Tracks AI Kill Switch and Independent Frontier-Model Audits in California

Overview

Governor Gavin Newsom signed an executive order on September 18 directing state agencies to accelerate implementation of two recently signed California AI laws and to convene outside experts on the possibility of a frontier-model “kill switch,” according to the governor’s office. The order comes as, according to the governor’s office, “the federal government has taken no action” on AI safety, with no federal law requiring AI companies to report dangerous incidents.

What We Know

The executive order accelerates implementation of Senate Bill 813, sponsored by state Sen. Jerry McNerney, which establishes a framework for independent verification organizations, and Assembly Bill 1405, sponsored by Assemblymember Rebecca Bauer-Kahan, which creates a state registry for AI auditors, according to the governor’s office. As Reason reports, the two bills create a state framework for certifying and registering independent AI auditors but do not require developers to submit to audits.

The order also convenes a panel of outside experts to recommend, within two months, ways California could strengthen its AI safety and security laws, according to the governor’s office. According to KALW, the panel’s recommendations are due November 16. Proposals under consideration include requiring independent third parties to write safety plans for frontier AI companies and requiring companies to develop an emergency shutoff, or “kill switch,” for frontier models, the governor’s office said. Reason reports the panel will also examine expanding the definition of “critical safety incidents” to include loss-of-control events such as the Hugging Face attack — a reference to the breach in which an autonomous AI agent executed more than 17,000 actions against Hugging Face’s systems over a weekend, as previously reported.

Speaking about the kill-switch proposal in an Instagram video, Newsom said the order is “advancing efforts to create an emergency shut-off, otherwise known as an AI kill switch, which means a lot of things depending on who you talk to,” adding, “so we want to flesh out exactly what that means and how to come up with a framework that works,” according to KALW. The governor’s office quoted Newsom saying, “We’re not waiting to act – we’re going to speed up our work on substantial and responsible AI oversight,” according to the governor’s office. In a Facebook post reported by Reason, Newsom wrote, “With Donald Trump and Congress failing to lead, California is stepping up to keep all Americans safe.”

The order builds on SB 53, the Transparency in Frontier Artificial Intelligence Act signed in 2025, and on Newsom’s first-in-the-nation generative AI executive order from 2023, according to the governor’s office. The move follows a similar step in Illinois, where Governor JB Pritzker issued an executive order creating an Illinois Artificial Intelligence Cabinet around the same time, according to Transparency Coalition.

Reactions

Not every observer is convinced the order changes much in practice. Adam Kovacevich, who heads the center-left tech-industry group Chamber of Progress, told KALW that “what Governor Newsom is mostly doing is trying to show that he’s jumping on the issue of the moment by saying something, but ultimately it doesn’t really do very much. It’s not actually creating a kill switch for the internet. It’s asking a group of experts to explore that issue as well as others.” Kovacevich also argued that “even though AI is raising in salience as a political issue, it’s still not in the top 10 of voters’ concerns. Voters are still primarily concerned with cost of living, healthcare, crime, education,” according to KALW.

Adam Thierer, a senior fellow at the R Street Institute, pushed back on the governor’s office describing California’s approach as a model for the country, telling Reason, “It’s not a guidebook. It’s a rule book.” Thierer warned that a patchwork of state AI regulations, rather than a consistent federal framework, risks “a complete inversion of the model that has governed the internet and digital commerce and speech for the last 30 years,” and said California “is once again attempting to position itself as a super regulator for the entire nation… in this case, they’re looking to be the overlord of all things AI,” according to Reason.

The order also revives a debate Newsom himself waded into in 2024, when he vetoed Senate Bill 1047 — legislation that would have created a state entity to oversee frontier models and required mandatory third-party audits, safety testing, and kill-switch capabilities, according to Reason. In his veto statement at the time, Newsom warned the bill could give the public a “false sense of security” by focusing on large-scale models while smaller models could pose similar risks, and cautioned, “if we over-regulate, if we overindulge, if we chase a shiny object, we could put ourselves in a perilous position,” Reason reported.

What We Don’t Know

The expert panel has not yet issued its recommendations, which are due November 16, according to KALW, and it remains unclear how any resulting kill-switch or audit requirements would be defined or enforced. Newsom is term-limited, and KALW noted that much of what happens next “will fall to Newsom’s successor,” according to KALW.