Anthropic Rejects Pentagon’s New Terms, Stands Firm on Lethal Autonomous Weapons and Mass Surveillance

CEO Dario Amodei issued a statement just before the Pentagon’s deadline, refusing the Department of Defense’s demand for unrestricted access to Anthropic’s AI. This marks the peak of an intense negotiation tied to Defense Secretary Pete Hegseth’s ambition to change AI lab contracts with the military. Anthropic maintains its stance against mass surveillance and autonomous lethal weapons, despite other companies like OpenAI agreeing to the terms. Amodei met with Hegseth at the White House, where an ultimatum was given. In Amodei’s late Thursday statement, he emphasized AI’s importance for national defense but warned against using AI for surveillance or fully autonomous weapons. He indicates that such technology isn’t reliable enough yet for defense. The Pentagon is reportedly assessing contractors’ reliance on Anthropic, potentially labeling it a “supply chain risk,” and may use the Defense Production Act to enforce compliance. Amodei, however, states that these potential actions won’t alter Anthropic’s stance and promises a transition to another provider if necessary, ensuring no disruption to military operations.