Overview
- Pentagon spokesperson Sean Parnell confirmed the department’s relationship with Anthropic is under review after disputes over usage limits on the Claude AI model.
- The Defense Department has threatened a supply‑chain risk label, a move typically reserved for foreign adversaries that could bar Anthropic from government work and pressure prime contractors to cut ties.
- At issue is the Pentagon’s push to allow “any lawful use” of AI, which collides with Anthropic’s red lines against mass domestic surveillance and fully autonomous weapons.
- Reports say Claude was used via Palantir in the classified operation to capture Nicolás Maduro, though Anthropic has neither confirmed nor detailed any specific operational use.
- Rival labs have accepted broader terms for some government work while Anthropic, which holds a DoD prototype contract worth up to $200 million and unique access on classified networks through partners, says talks with the department continue.