Anthropic launches reporting tools to track enterprise AI spending
The company is also creating shared team budgets as businesses seek a clearer view of AI coding costs and results.
Anthropic urges California to let courts halt risky AI releases
In a letter to Governor Gavin Newsom, the company also opposed mandatory model-wide shutdown controls, favoring narrower interventions.
Anthropic has urged California Governor Gavin Newsom to support a process allowing officials to seek judicial orders preventing the release of artificial intelligence models found to present serious public risks. The company outlined its position in an October 15 letter.
The recommendations followed Newsom's decision in September to assemble AI experts to propose safety requirements. An Anthropic spokesperson verified the letter's authenticity; the governor's office did not immediately provide a comment.
Anthropic opposed requiring developers to install controls that could shut down a whole model if the company lost control of its actions. It argued that narrower measures, including disabling an individual AI agent or ending access for a particular customer, would work better.
The company has itself been subject to government limits. In June, the Trump administration instructed Anthropic to immediately curb overseas access to its cyber-capable Mythos 5 and Fable 5 models on national security grounds. The Commerce Department subsequently removed that restriction.
Federal lawmakers have also considered intervention powers. According to a media report, a bipartisan Senate group debated a proposal in September allowing the government to prevent certain models deemed unsafe from being released, with companies able to contest decisions in federal court. The proposal has not advanced since Congress departed for its pre-election recess.
Separately, House lawmakers proposed legislation in July called the AI Kill Switch Act. It would authorize officials to require AI companies to shut down models that threaten human life or the economy.