Pentagon Memo Orders Removal of Anthropic AI from National Security Systems

TLDR

A March 6 internal DOD memo directs the removal of Anthropic technologies from key national security and operational systems. The order follows the earlier supply chain risk label and comes as Anthropic challenges the designation in court.

Details of the Directive

The memo, addressed to senior military leaders and commanders, requires agencies to identify and discontinue use of Anthropic models and services. It cites national security concerns over the company's refusal to relax safeguards for certain military applications.

Implementation timelines vary by system criticality, with some immediate actions required. Reports indicate the memo references Anthropic's Claude being used in sensitive operations, including aspects of the Iran campaign.

Anthropic's Position

The company maintains that its safeguards are essential and that the government's demands would compromise core principles. It has reiterated that most commercial customers are unaffected and has filed lawsuits challenging the designation as retaliatory.

Why this story matters

The directive represents a significant escalation in the U.S. government's use of administrative tools against a domestic AI developer. It tests the boundaries of how far national security priorities can override commercial and ethical stances in the AI sector.

Sources

• Internal DOD memo dated March 6, 2026 (obtained and reported by CBS News, The Hill)
• Anthropic statements on the ongoing dispute (anthropic.com/news, early March 2026)
• Coverage in The New York Times, Reuters, and defense-focused publications
• Related Senate letters and reporting on the designation timeline

Featured Image Alt Text

Pentagon building with Anthropic logo and warning overlay representing the removal directive

Tags

Pentagon, Anthropic, AI Policy, National Security, Supply Chain Risk, Memo