Anthropic has told a US appeals court that it cannot manipulate its artificial intelligence tool Claude once deployed in classified Pentagon military networks, aiming to counter the Trump administration's attempt to label the company a supply chain risk.
The statement, part of a 96-page filing with the US Court of Appeals in Washington DC, provides insight into arguments Anthropic's lawyers plan to make in a lawsuit filed last month over a contract dispute regarding the use of AI in fully autonomous weapons and potential surveillance of Americans.
San Francisco-based Anthropic alleges the Pentagon is illegally retaliating by stigmatising it with a designation meant to protect national security systems from foreign sabotage. Earlier this month, the appeals court rejected Anthropic's request to block the Pentagon's actions while evidence is collected.
Anthropic's new filing addresses the court's questions ahead of oral arguments scheduled for May 19. The Trump administration will file its response before the hearing. This setback follows Anthropic's victory in a separate case in San Francisco federal court, which prompted the administration to remove the labels from Anthropic.
The lack of a similar order in the Washington case continues to affect Anthropic, whose AI tools have made it a rising tech star alongside OpenAI. After the Pentagon cancelled a $200 million contract with Anthropic, OpenAI struck a deal to provide its technology to the US military.
