Anthropic’s Deal With US Military Under Threat

The government is also “close” to designating the company a “supply chain risk,” which means anyone wishing to do business with the military would also need to end their relationship with Anthropic, Axios reported Monday (Feb. 16), citing a senior Pentagon official.

“It will be an enormous pain in the ass to disentangle, and we are going to make sure they pay a price for forcing our hand like this,” that official said.

The report noted that this sort of penalty is typically reserved for foreign adversaries. And Anthropic’s Claude is the only AI model available in the military’s classified systems, with the Pentagon championing its abilities, Axios added.

But for the last few months, the report said, Anthropic and the Pentagon have been holding contentious talks on the military’s permitted usage of Claude. The company is prepared to relax its terms of use, but wants assurances its tech won’t be used for mass domestic spying operations or to build weapons that can be deployed with no human involvement.

According to Axios, the Pentagon says this is too restrictive, with a host of gray areas that would make such terms impractical. The military is insisting in negotiations with Anthropic and other major AI firms that it be permitted to use their tools for “all lawful purposes,” the report added.

“The Department of War’s relationship with Anthropic is being reviewed. Our nation requires that our partners be willing to help our warfighters win in any fight,” Chief Pentagon spokesman Sean Parnell told Axios. “Ultimately, this is about our troops and the safety of the American people.”

An Anthropic spokesperson told Axios: “We are having productive conversations, in good faith, with DoW on how to continue that work and get these new and complex issues right.”

The Pentagon last year awarded up to $200 million each to four U.S.-based AI companies developing “frontier” models—Anthropic, Google, OpenAI and xAI—as the military’s Chief Digital and Artificial Intelligence Office (CDAO) looks to employ agentic AI to deal with national security challenges.

CDAO Head Doug Matty said the goal is to tap into the best technologies developed by American AI firms to support its troops and uphold a strategic advantage.

“The U.S. military is no slouch when it comes to technological innovation,” PYMNTS wrote at the time. “For example, the DoD’s R&D arm — Defense Advanced Research Projects Agency or DARPA — created Arpanet in 1969, a communications network that linked computers far apart. Arpanet later became the internet.”