Pentagon Excludes Anthropic from Classified AI Deals Over Safety Concerns

Summary

The US Department of Defense on Friday signed agreements with seven major AI companies—including OpenAI, Google, Microsoft, Amazon Web Services, NVIDIA, SpaceX, and Reflection—to deploy their AI systems on secure classified Pentagon networks. Notably absent from the deals is Anthropic, the maker of Claude, which was excluded due to a dispute over safety restrictions on military applications. The Pentagon and its contractors had previously labeled Anthropic as a supply-chain risk, citing concerns about the company's approach to AI safety. The integration of these AI systems into classified networks (Impact Levels 6-7) will provide over 1.3 million Defense Department personnel with access to advanced AI tools for operational use.

The exclusion marks a significant setback for Anthropic, as it means Claude will be replaced by ChatGPT and competing systems on the military's most sensitive networks. Defense Department Chief Technology Officer Emil Michael confirmed that Anthropic remains a supply-chain risk, though he separately described Anthropic's Mythos AI model as a potential "national security moment" due to its advanced cyber capabilities. However, recent diplomatic progress suggests the landscape may be shifting: Anthropic CEO Dario Amodei met with senior Trump administration officials at the White House earlier this month to discuss the company's work, with both sides calling the meeting productive. President Trump indicated that Anthropic is "shaping up" in his administration's eyes, raising the possibility that the company's Pentagon blacklisting could eventually be reversed.

• Anthropic CEO recently met with Trump administration officials; reversal of Pentagon ban remains possible