Governor Newsom vetoes California SB 1047 and convenes the Joint California Policy Working Group on AI Frontier Models (29 September 2024)

On Sunday 29 September 2024, Governor Gavin Newsom returned California Senate Bill 1047 — the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, authored by State Senator Scott Wiener and co-sponsored by Encode Justice, the Center for AI Safety Action Fund, and Economic Security Action California — to the California State Senate without his signature. The veto closed the first substantive US-state legislative attempt at frontier-AI safety regulation. Newsom paired the veto with an announcement the same day that he would convene a Joint California Policy Working Group on AI Frontier Models led by Stanford's Fei-Fei Li, Carnegie Endowment president Mariano-Florentino Cuéllar, and UC Berkeley's Jennifer Tour Chayes. The veto and the paired working-group announcement together constitute the closing event of the Encode Justice SB 1047 co-sponsorship campaign, and the opening of the policy process whose 17 June 2025 California Report on Frontier AI Policy would supply much of the substantive material for Wiener's narrower successor bill SB 53.

Context

SB 1047 had been introduced on 7 February 2024 and had moved through both chambers over the spring and summer with Encode Justice as the lead civil-society co-sponsor. The California Senate had passed the bill 32–1 on 21 May; the Assembly had passed the amended bill 48–16 on 28 August; and the Senate had concurred 30–9 on 29 August, sending the bill to Newsom on a thirty-day signing clock that ran to 30 September 2024.

The veto-eve mobilisation in September 2024 had assembled the broadest civil-society coalition any US-state frontier-AI safety bill had then put on the public record: a 9 September letter to Newsom from SAG-AFTRA, a same-day letter signed by more than 113 current and former employees of OpenAI, Google DeepMind, Anthropic, Meta, and xAI, and the 25 September Artists 4 Safe AI open letter that the Encode Justice campaign coordinated and that gathered signatures from more than 125 Hollywood actors, directors, and producers. Opposition through summer had been led by Google, Meta, OpenAI, Andreessen Horowitz, and Y Combinator, with public criticism from Speaker Emerita Nancy Pelosi, Representative Zoe Lofgren, and Stanford computer scientist Fei-Fei Li.

What happened on the day

Newsom released his veto message from the Office of the Governor in Sacramento, dated SEP 2 9 2024, returning the bill to the California State Senate without his signature. He paired the veto with a separately released gubernatorial announcement, titled "Governor Newsom announces new initiatives to advance safe and responsible AI, protect Californians", that named three academic experts to lead a new working group on frontier-AI policy and confirmed that he had signed seventeen AI-related bills in the preceding thirty days, including SB 896 directing the California Office of Emergency Services to assess generative-AI risks to the state's critical infrastructure.

The veto message

In the veto message, Newsom described SB 1047 as a bill that "would require developers of large artificial intelligence (AI) models, and those providing the computing power to train such models, to put certain safeguards and policies in place to prevent catastrophic harm" and that would also have "establish[ed] the Board of Frontier Models — a state entity — to oversee the development of these models". California, he wrote, "is home to 32 of the world's 50 leading AI companies", and the state had a responsibility to regulate the industry — a responsibility he said he took seriously.

The substantive criticism in the message turned on the bill's threshold for coverage. By focusing only on the most expensive and large-scale models, Newsom wrote, "SB 1047 establishes a regulatory framework that could give the public a false sense of security about controlling this fast-moving technology. Smaller, specialized models may emerge as equally or even more dangerous than the models targeted by SB 1047." The bill, he said, "does not take into account whether an AI system is deployed in high-risk environments, involves critical decision-making or the use of sensitive data" — "instead, the bill applies stringent standards to even the most basic functions — so long as a large system deploys it. I do not believe this is the best approach to protecting the public from real threats posed by the technology."

Newsom paired the criticism with explicit agreement on the underlying threat assessment. "Let me be clear — I agree with the author", he wrote, "we cannot afford to wait for a major catastrophe to occur before taking action to protect the public. California will not abandon its responsibility. Safety protocols must be adopted. Proactive guardrails should be implemented, and severe consequences for bad actors must be clear and enforceable." His objection, on the message's own terms, was not to the principle of state regulation of frontier AI but to a framework "not informed by an empirical trajectory analysis of AI systems and capabilities". Any framework for regulating AI, he concluded, "needs to keep pace with the technology itself".

The paired working-group announcement

The 29 September 2024 gubernatorial announcement named three experts to lead what would become the Joint California Policy Working Group on AI Frontier Models: Dr. Fei-Fei Li, co-director of the Stanford Institute for Human-Centered Artificial Intelligence; Mariano-Florentino "Tino" Cuéllar, president of the Carnegie Endowment for International Peace and member of the National Academy of Sciences Committee on Social and Ethical Implications of Computing Research; and Jennifer Tour Chayes, dean of the College of Computing, Data Science, and Society at UC Berkeley. The announcement framed the group's mandate as helping California "develop workable guardrails for deploying GenAI, focusing on developing an empirical, science-based trajectory analysis of frontier models and their capabilities and attendant risks" — the same "empirical trajectory analysis" language the veto message had used as its diagnostic test for what SB 1047 lacked.

The pairing — veto plus working group on the same day — converted a binary signing decision into a process commitment. It is the procedural manoeuvre against which subsequent state-AI-policy organising elsewhere in the corpus has been read.

Immediate reactions

Sneha Revanur, Encode Justice's founder, released a same-day reaction statement calling the veto "painful and disappointing" and saying that "it sends a signal that the tech lobby's chokehold is far tighter than the people's". She cited polling showing 77% of Californians supported the bill and framed the veto, on the record, as "an opening salvo. Next time, our movement will win. In some ways, it already has — we've catapulted the conversation about AI policy into the spotlight." A subsequent Stanford Daily retrospective recorded Revanur's reflective framing on the campaign's coalition-building — "the outcome is not what we'd hoped for", but "to see how far we came, how many allies we activated" — and confirmed Encode Justice's coordinating role behind the Artists 4 Safe AI open letter.

Senator Scott Wiener, the bill's author, called the veto "a missed opportunity for California to once again lead on innovative tech regulation" in a same-day statement carried by CalMatters. Nathan Calvin, the Center for AI Safety Action Fund's state legal lead and Encode Justice's working partner on the Sacramento track, would in subsequent weeks frame the campaign's substantive defeat as a building-block for the working-group process and for the successor legislation that Encode and Wiener immediately began drafting.

Significance

The 29 September 2024 veto closes the Encode Justice SB 1047 co-sponsorship campaign on its substantive merits — the bill did not become law — and is the corpus's first state-legislative-decision Event. Where the Andersen v. Stability AI class-action filing sits at the opening edge of one creative-industry track in the events slice, the 29 September 2024 veto sits at the closing edge of the corpus's first US-state frontier-AI-safety track: the moment at which a youth-led civil-society coalition had assembled the broadest coalition any such bill would assemble — Hollywood actors, SAG-AFTRA, AI-lab employees, the "Godfathers of AI" Geoffrey Hinton and Yoshua Bengio, Anthropic CEO Dario Amodei's qualified support — and still lost the signing decision.

The pairing with the working-group announcement gives the day a second analytical handle. The Joint California Policy Working Group on AI Frontier Models would, on 17 June 2025, publish a 53-page California Report on Frontier AI Policy whose recommendations on third-party evaluation, incident reporting, whistleblower protection, and risk disclosure carried forward several of SB 1047's load-bearing provisions. Wiener and Encode Justice would draft those recommendations into the narrower successor bill SB 53, the Transparency in Frontier Artificial Intelligence Act, which Newsom signed into law on 29 September 2025 — exactly one year after the SB 1047 veto, making SB 53 the first US statute focused squarely on frontier-AI safety. Read against that one-year arc, the 29 September 2024 veto is legible in retrospect not as the campaign's endpoint but as the inflection between its bill-passage phase and its working-group-and-successor-bill phase — the closing-counterparty bookend that the queued opening-counterparty messaging and voice entries (msg-existential-threat-to-creative-professions, pub-artists-4-safe-ai-letter, voice-karla-ortiz) will continue to fill out from the civil-society side.

For the wider make-AI-good movement this corpus is mapping, the veto matters on three further counts. First, it is the first US-state-executive decision on a frontier-AI safety statute on the public record, and the proximate cause of the only state-level working-group process — the California cafrontieraigov.org track — to have produced a substantive cross-discipline policy report in 2025. Second, the veto message's "empirical trajectory analysis" framing has been adopted explicitly by subsequent state-level AI-safety bills and by federal AI-policy discussions, becoming the live rhetorical alternative to compute-threshold regulation that civil-society co-sponsors must now answer to. Third, the campaign's youth-led coalition template — Sacramento inside work, Hollywood public-letter mobilisation, AI-researcher validator endorsements, AI-lab-employee solidarity — survived the bill's defeat and has been carried by Encode Justice into the SB 53 process and into successor state campaigns elsewhere on the corpus's US-state-policy front.