Microsoft and a coalition of former military leaders and AI think tanks are backing Anthropic in its legal bid to overturn the Trump administration's "supply chain risk" designation. Microsoft argues the designation imposes vague, ill-defined directions on government contractors, potentially causing severe economic harm and acting against the public interest.
Anthropic's challenge stems from the US Department of War's (DOW) ability to exclude the company from contract awards, remove its products, and prevent its use by prime contractors. Until recently, Anthropic was notably approved for classified military networks, with its AI chatbot deployed across government systems, including national nuclear laboratories and for intelligence analysis.
Joint filings supporting Anthropic have come from prominent American military officials, including Michael Hayden, former CIA director, who stated US Defense Secretary Pete Hegseth's actions "threaten the rule-of-law principles." These filings allege a misuse of government authority for retribution. Ex-military officials warn that targeting widely embedded military technology creates "sudden uncertainty" and risks soldier safety.
Furthermore, 37 AI engineers from former OpenAI and Google DeepMind roles have filed a brief calling the DOW's actions an "improper and arbitrary use of power" with serious ramifications for U.S. industrial and scientific competitiveness in AI. They caution that such actions could chill open deliberation on AI risks and benefits.
The conflict began when Anthropic refused to grant the military unfettered access to its AI chatbot, Claude. Anthropic CEO Dario Amodei cited two key restrictions: that its technology not be used for mass domestic surveillance or embedded in fully autonomous weapons. Amodei stated he "cannot in good conscience accede to the Pentagon's request" for unrestricted access, believing some AI uses undermine democratic values or exceed current safe and reliable capabilities.
Microsoft's filing echoed these concerns, asserting that "American AI should not be used to conduct domestic mass surveillance or start a war without human control." Claude is slated for phasing out of military operations in the next six months.