Anthropic CEO Dario Amodei announced that the company “cannot in good conscience accede” to the Pentagon’s demands for broader use of its AI technology. Although the AI chatbot maker Claude has not withdrawn from talks, Amodei criticized the Defense Department’s revised contract terms for failing to restrict mass surveillance or fully autonomous weapons.
Pentagon spokesman Sean Parnell asserted the military’s intent to use Anthropic AI legally, denying plans for mass surveillance or autonomous weapons. Parnell emphasized that no company will dictate military operational decisions.
During a recent meeting, Defense Secretary Pete Hegseth mentioned potential actions against Anthropic, including contract termination, being labeled a supply chain risk, or invoking the Defense Production Act to compel product use. Amodei argued these actions are contradictory, labeling the company both a risk and vital to national security.
Separately, Senators Thom Tillis, a North Carolina Republican, and Mark Warner, a Virginia Democrat, expressed concerns about the Pentagon’s approach. Tillis criticized public negotiations, while Warner highlighted the need for AI governance. Talks intensified after months of discussions, with Anthropic open to a transition should an agreement fail.
