OpenAI Signs Pentagon Classified Systems Deal

On February 27, 2026, hours after the Trump administration designated Anthropic a supply chain risk and ordered all federal agencies to stop using its products, OpenAI CEO Sam Altman announced his company had reached an agreement with the Department of Defense to deploy OpenAI models on classified military networks. The timing was not coincidental — the deal positioned OpenAI as Anthropic’s direct replacement on the Pentagon’s most sensitive AI infrastructure.

Anthropic’s Claude had been the only AI model operating on classified Pentagon networks, deployed through Palantir’s integration layer. OpenAI’s models would now fill that role. In a memo to employees, Altman claimed OpenAI shared Anthropic’s “red lines” on military AI use — a statement that created an immediate and unresolved contradiction. If both companies held the same safety positions, the question of why the Pentagon accepted OpenAI’s terms while blacklisting Anthropic had no obvious answer beyond selective enforcement.

The speed of the announcement revealed advance preparation. Deploying AI models on classified networks requires extensive security reviews, infrastructure integration, and personnel clearances that cannot be arranged in hours. The deal had clearly been negotiated in parallel with the Anthropic confrontation, ready to announce the moment Anthropic was removed. This sequence established the Pentagon’s strategy: Anthropic’s safety stance was not a negotiation to be resolved but an obstacle to be routed around.

Industry analysts described the move as a signal to the entire AI sector. No individual AI company had irreplaceable leverage. Companies that resisted government demands on safety guardrails would be replaced — immediately — by companies that complied. The incentive structure was clear, and the message was received.