Anthropic tells court it has no control over AI once deployed in Pentagon classified networks

Anthropic seeks to debunk Pentagon's claims about its control over AI technology in military systems

WASHINGTON (AP) -- Anthropic on Wednesday told an appeals court that it can't manipulate its artificial intelligence tool Claude once it is deployed in classified Pentagon military networks -- an assertion aimed at debunking the Trump administration's attempt to brand the rapidly growing technology company as a supply chain risk. The statement made as part of 96-page filing with the U.S. Court of Appeals in Washington D.C. provided a glimpse at the arguments that Anthropic's lawyers intend to make as part of a lawsuit filed last month in the fallout of a contract dispute over how AI technology can be used in fully autonomous weapons and potential surveillance of Americans. San Francisco-based Anthropic contends the Pentagon is illegally retaliating against it by stigmatizing it with a designation meant to protect against sabotage of national security systems by foreign adversaries. Earlier this month, the appeals court rejected Anthropic's request for an order that would have blocked the Pentagon's actions while the panel is still collecting evidence about the case. Anthropic's new filing is meant to directly address some of the court's questions ahead of oral arguments scheduled for May 19. The Trump administration will have an opportunity to file its response before that hearing. Anthropic's temporary setback in the Washington case came after it already had prevailed in a separate case focused on the same issues in San Francisco federal court. That decision prompted the Trump administration to remove the stigmatizing labels from Anthropic, according court filings. But the lack of a similar order in the parallel case in Washington continues to cast a cloud over Anthropic, whose AI tools have turned it into a rising tech star along with rival OpenAI. After the Pentagon canceled a $200 million contract with Anthropic in the wake of their disagreement, OpenAI struck a deal to provide its technology to the U.S. military.

Anthropic: No "kill switch" for AI in classified settings

Why it matters: The Pentagon designated Anthropic a supply chain risk, contending the AI firm is inappropriately getting involved in how its technology can be used in sensitive military operations. What's inside: Anthropic argues in the filing to a federal appeals court in D.C. that it has no visibility, technical ability or any kind of "kill switch" for its technology once it's deployed. * The company also says the Pentagon has the opportunity to test models before deployment. Catch up quick: The company's usage policies include no Claude for autonomous weapons or mass surveillance, red lines that the Pentagon dismissed as red herrings and led to the dispute. * The D.C. court previously rejected Anthropic's request for a pause on the supply chain risk designation. A judge in California for an ongoing parallel case granted Anthropic's request. * The split decision means Anthropic can't participate in new Pentagon contracts, but can continue working with other government agencies while the litigation plays out. Friction point: The Pentagon is arguing in court that Anthropic is a supply chain risk as the Trump administration moves to deploy its new Mythos model across the federal government. * Now, agency heads are scrambling to figure out how they can protect their systems from cyber attacks using Mythos, potentially complicating the administration's argument that the company poses a national security risk. What's next: A hearing is scheduled for May 19.