Particle.news
Download on the App Store

Pentagon Designates Anthropic a Supply-Chain Risk Effective Immediately

The unprecedented move against a domestic AI vendor tests the Pentagon’s demand for unrestricted use of commercial models.

Overview

  • Anthropic said it will challenge the designation in court and emphasized that the restriction applies to defense work, with Microsoft noting Claude can still be used on non-Defense projects and platforms.
  • Despite the label, reporting indicates the Pentagon continued using Claude to support operations, including in Iran, creating near-term transition and mission-continuity risks.
  • CEO Dario Amodei apologized for a leaked memo criticizing the Trump administration and OpenAI, and multiple outlets report he has re-engaged with Under Secretary Emil Michael, though a senior official publicly denied active negotiations.
  • The rupture stems from Anthropic’s refusal to drop contractual limits on mass domestic surveillance and fully autonomous weapons, clashing with the Pentagon’s insistence on access for all lawful purposes.
  • Lawmakers and former national-security officials criticized using a supply-chain authority typically aimed at adversary technology, defense contractors began shifting away from Anthropic, and OpenAI’s separate deal drew scrutiny before it added explicit guardrails on domestic surveillance.