Artificial intelligence firm Anthropic has called on California Governor Gavin Newsom to grant state officials the authority to petition courts for injunctions against AI models that present severe public dangers. This suggestion arrives as lawmakers across the United States weigh tighter regulations for sophisticated AI technologies and strategies to avert potential hazards.
In an October 15 correspondence, the organization argued that authorities ought to have intervention powers whenever safety assessments flag a model as a major hazard. These suggestions from Anthropic could shape California’s regulatory strategy for artificial intelligence, potentially influencing legislative conversations nationwide.
Anthropic Calls for Stronger AI Safety Oversight
The company put forward its proposals following a September panel convened by Governor Newsom that brought together artificial intelligence specialists to formulate safety standards. Anthropic backs independent safety audits and consistent reporting regarding the hazards presented by advanced AI models.
Under this framework, outside auditors would have the ability to transmit critical warnings straight to the California attorney general. Should authorities detect an imminent danger of catastrophic damage, they should possess the capability to request a judicial order halting a model’s release, internal deployment, or ongoing development.
Furthermore, Anthropic advocates for the publication of risk assessments at least once every half-year for major AI creators. The firm contends that third-party oversight would assist regulators in catching potential issues prior to widespread harm occurring.
Company Opposes Blanket AI Kill Switch Rules
While endorsing emergency controls for artificial intelligence platforms, Anthropic has expressed reservations regarding mandates that compel developers to engineer systems featuring total shut-down mechanisms.
The business advocates for precise interventions instead, such as deactivating a malfunctioning autonomous agent or limiting a specific client’s access. The company maintains that such steps can resolve isolated issues effectively without forcing a complete system shutdown.
According to the firm, full system shutdowns ought to be kept for the most critical scenarios. Its strategy prioritizes ongoing observation of AI systems and localized responses to emerging issues.
Also Read: Anthropic IPO: What We Know About Its Planned Nasdaq Listing
AI Regulation Debate Gains Momentum
The recommendation emerges amidst ongoing discussions in Congress regarding whether the federal government should possess the power to halt the deployment of hazardous AI models. A bipartisan congressional group reviewed similar legislation in September, though the measure stalled prior to the pre-election recess in Congress.
Additionally, the US House evaluated the AI Kill Switch Act, which proposes granting authorities the right to mandate the deactivation of AI systems deemed dangerous to human life or economic stability.
Anthropic’s proposals contribute to the broader conversation concerning how governments can reconcile technological innovation with public protection. Policymakers face the core challenge of defining transparent criteria for regulatory intervention without unduly hindering the creation of beneficial AI innovations.


