Anthropic CEO Dario Amodei has taken a rare stand against the US Department of Defense, refusing to strip safety safeguards from the company’s Claude AI model for specific military applications. Amodei drew explicit red lines against using the technology for mass domestic surveillance or fully autonomous lethal weapons, citing significant reliability risks.
The Pentagon retaliated in March by designating Anthropic a supply-chain risk, a move that threatens to exclude the firm from federal contracts. This marks the first time such a label has been applied to a US AI company. Anthropic subsequently filed a lawsuit, arguing the designation is legally unsound and that maintaining safety protocols should not constitute a national security liability.
By May 1, the Department of Defense expanded classified AI contracts to eight other firms, explicitly excluding Anthropic. Competitors like OpenAI and Microsoft are positioned to fill the void. Despite the standoff, limited government use of Claude remains permitted, and discussions with White House officials continue as the Trump administration pushes for accelerated AI adoption.