OpenAI will deploy its AI models inside the Pentagon’s classified network. The deal lands hours after the Defense Department labeled Anthropic a supply-chain risk, detonating one of the government’s most prominent AI vendor relationships.
Sam Altman said late Friday the agreement reflects OpenAI’s principles, including prohibiting “domestic mass surveillance” and keeping “human responsibility for the use of force” in autonomous weapon systems. He said OpenAI built safeguards for the classified deployment and that the tools will run “only” on cloud networks, not edge servers. The Pentagon did not respond to requests for comment.
Tonight, we reached an agreement with the Department of War to deploy our models in their classified network.
— Sam Altman (@sama) February 28, 2026
In all of our interactions, the DoW displayed a deep respect for safety and a desire to partner to achieve the best possible outcome.
AI safety and wide distribution of…
The episode hardens a split among frontier-model vendors over military constraints. Anthropic has said its products should not be used for surveillance of Americans or to make fully autonomous weapons, and it framed the Pentagon move as punishment for that stance. OpenAI, which once barred military use, updated its policy in 2024 to allow such applications, while positioning controls as enforceable terms inside deployments.
Money and market power sit behind the policy language. OpenAI said Friday it raised $110 billion at a $730 billion valuation, while Anthropic raised $30 billion earlier this month, according to the report. Both firms are pushing enterprise adoption and have signaled interest in public listings as soon as this year, making government contracts and reputational risk harder to separate.
Altman urged the department “to offer these same terms to all AI companies.” Defense Secretary Pete Hegseth said Anthropic has six months to hand over AI services to another provider and wrote, “This decision is final.” Anthropic said “no amount of intimidation or punishment” would change its position and pledged to challenge any formal designation in court.
OpenAI’s foothold inside classified cloud systems now sits alongside other Pentagon-linked AI work, including an effort using models to translate voice commands into digital instructions for autonomous drone swarming, according to a document reviewed by Bloomberg. Will the Pentagon’s six-month transition window become the template for how frontier models enter classified workflows? The next catalyst is whether Anthropic’s supply-chain label is formalized and litigated, while OpenAI’s cloud-only deployment begins operating at scale.
