US Media: Pentagon Threatens to Sever AI Collaboration with Anthropic

BlockBeats News, February 15 — According to Axios, a senior government official revealed that the U.S. Department of Defense is considering ending its partnership with Anthropic because the AI company insists on certain restrictions on military use of its models. It is reported that the Pentagon is urging four leading AI laboratories to allow the military to use their tools for “all lawful purposes,” including weapons development, intelligence gathering, and battlefield operations.

Anthropic has not accepted these terms, and after months of difficult negotiations, the Pentagon has grown frustrated. Anthropic maintains that two areas must remain prohibited: mass surveillance of the American public and fully autonomous weapons systems. The senior official stated that there is considerable ambiguity about which situations should fall into these categories and which should not.

Furthermore, it is not feasible for the Pentagon to negotiate each specific use case with Anthropic or for Claude to unexpectedly block certain applications. The official said, “Anything is possible,” including temporarily reducing cooperation with Anthropic or completely ending the partnership. “But if we believe that is the right course, then we must find suitable replacements for them.” (Jin10)