Anthropic has told California Governor Gavin Newsom that government officials should be able to seek court orders blocking the release of AI models that pose a “serious risk” to the public. The position was set out in an October 15 letter, confirmed as authentic by an Anthropic spokesperson, and first detailed by Politico.
The letter follows Newsom’s September decision to convene a panel of AI experts to develop proposals for AI safety requirements. OpenAI sent its own list of recommendations to the governor on October 20, underlining how the biggest labs are now actively shaping the rules that could govern their future releases.
Anthropic’s argument is narrower than a blanket shutdown power. The company warned that requiring developers to build “kill switches” capable of shutting down entire models could give officials a false sense of reassurance. Instead, it favors targeted interventions, such as disabling a problematic AI agent or cutting off a specific customer’s access, which it says would be more effective in practice.
The proposal echoes debates already underway in Washington. A bipartisan group of senators discussed legislation in September that would let the government block certain unsafe models, with companies able to challenge the decision in federal court, while House lawmakers proposed an “AI Kill Switch Act” in July. Neither proposal has advanced, and Congress has since left for its pre-election break.
Anthropic itself has direct experience of government intervention: the Trump administration ordered it in June to restrict foreign access to its cyber-capable Mythos 5 and Fable 5 models on national security grounds, before the Commerce Department later lifted the order. For users of Claude, ChatGPT, and Gemini, the California debate is a sign that safety rules for powerful models are moving from theory toward law.
Source: Reuters via Mix 92.9
