Why the Pentagon Wants to Destroy Anthropic
About this episode
Ezra Klein asks former White House AI adviser Dean Ball about the Pentagon's campaign against Anthropic after the Defense Department demanded military use of Claude without usage restrictions. A 2024 classified-use agreement allowed military use while barring domestic mass surveillance and fully autonomous lethal weapons; the Trump administration expanded it in 2025, then rejected the restrictions. When Anthropic refused to drop them, the Pentagon canceled the contract and threatened an unprecedented 'supply chain risk' designation against a U.S. company. Ball agrees elected government — not a CEO — should decide military uses, but calls the punitive designation corporate assassination likely beyond statutory authority. They discuss how AI turns commercially bought location and behavioral data into scalable surveillance, the clash between state sovereignty and private AI power, and whether private labs could be nationalized — agreeing identifiable humans must stay legally accountable for autonomous systems.