OpenAI announced a new agreement with the U.S. Department of Defense to run its AI models on the Pentagon’s classified networks, a move revealed hours after the White House ordered federal agencies to stop using technology from rival Anthropic.
In a late Friday post on X, OpenAI CEO Sam Altman said the company will host models inside the Defense Department’s classified environment and that Pentagon officials showed “deep respect for safety” and willingness to operate within OpenAI’s safety constraints.
The deal follows a rapid escalation this week. Defense Secretary Pete Hegseth designated Anthropic a “Supply-Chain Risk to National Security,” a label normally reserved for foreign adversaries. The designation requires defense contractors to attest they are not using Anthropic’s models. At the same time, President Donald Trump directed all federal agencies to immediately stop using Anthropic products, allowing a six-month transition for agencies already dependent on the company’s systems.
Anthropic had been the first lab to deploy AI models within the Pentagon’s classified environment under a $200 million contract signed in July. Talks between Anthropic and the Defense Department broke down after Anthropic sought assurances its models would not be used for autonomous weapons or domestic mass surveillance. The department pushed back, saying the technology must be available for all lawful military purposes.
Anthropic called the designation “deeply saddening” and said it will challenge the decision in court. The company warned the move could reshape how U.S. technology firms negotiate with government agencies amid growing political scrutiny of AI partnerships.
Altman said OpenAI’s agreement includes comparable restrictions, including a prohibition on domestic mass surveillance and clauses requiring human accountability for force-related decisions, even where automated systems are involved. According to OpenAI, those limits are built into the contract governing use inside the classified network.
The announcement prompted a mix of support, skepticism and criticism online. Some users criticized OpenAI for working with the Pentagon, citing earlier promises not to enable weapons or surveillance; others said they would migrate to Anthropic alternatives. One public figure posted that they had canceled ChatGPT and switched to Anthropic’s Claude Pro Max, while other commentators accused OpenAI of reversing earlier commitments.
This report was prepared in line with the publisher’s editorial standards and aims to provide an accurate, timely summary of recent developments. Readers are encouraged to verify details independently.