Pentagon Presses OpenAI, Anthropic to Expand AI on Classified DoD Networks

The U.S. Department of Defense is urging leading AI firms including OpenAI and Anthropic to make their tools available on classified networks with fewer standard restrictions. At a White House event on February 11, 2026, Pentagon Chief Technology Officer Emil Michael said the military aims to deploy frontier AI across all classification levels, including classified domains used for mission planning and targeting. Of the companies in active talks, OpenAI has agreed to make its tools available on the unclassified genai.mil network to over 3 million Defense Department employees, removing many user guardrails while maintaining legal and ethical boundaries. Anthropic is engaged in contentious discussions over classified access, with the firm opposing autonomous weapon targeting and domestic surveillance. The expansion seeks to leverage AI for battlefield information synthesis but raises risks of error and false information in sensitive settings. Anthropic and OpenAI have each sought to preserve safeguards and requested new authorizations for classified use beyond current agreements.