Pentagon Signs Defense AI Deals with 8 Companies; Anthropic Out
The US Department of Defense signed "any lawful purpose" defense AI agreements on May 1 with OpenAI, Google, xAI, Microsoft, NVIDIA, AWS, Oracle, and Reflection AI — granting access to classified networks. Anthropic was explicitly excluded after a prolonged dispute over ethical guardrails; the company continues to face Pentagon lawsuits and has lost government revenue as a result. Some contractors reportedly received informal non-lethal and surveillance-limit assurances from the DoD — essentially the guardrails Anthropic had demanded publicly.
Why It Matters
The deal formalizes the end of "AI companies won't touch defense." Anthropic holds the moral high ground but no contract, while OpenAI and Google now shape classified AI model use.