Overview
- The Defense Department labeled Anthropic a supply‑chain risk and directed federal agencies to stop using Claude with a six‑month phaseout, an uncommon move for a U.S. company.
- OpenAI amended its fresh Pentagon agreement after protests from users and employees, adding a clause barring intentional domestic surveillance and requiring a separate modification before intelligence‑agency use.
- Defense tech firms and major contractors are preemptively replacing Claude to stay compliant, raising short‑term disruption risks where the model had been embedded through partners such as Palantir.
- Anthropic says it will contest any formal designation in court and reiterates it will serve national security under its stated red lines on mass surveillance and fully autonomous weapons.
- Media outlets reported Claude supported recent U.S. strikes on Iran, but neither the Pentagon nor Anthropic confirmed the claim.