OpenAI signs pact with Pentagon to deploy AI models on classified network

Sam Altman says OpenAI will deploy models on the US Department of War's classified network under strict safeguards, barring domestic mass surveillance and ensuring human control over force use

Sam Altman on Saturday said OpenAI has reached an agreement with the US Department of War to deploy its artificial intelligence models within the department’s classified network. The move marks a significant step in the company’s engagement with national security institutions.

 

In a post on X, Altman said the department displayed “a deep respect for safety” and a desire to partner in achieving the best possible outcome. He stressed that AI safety and the wide distribution of benefits remain central to OpenAI’s mission.

 

Altman added that two of OpenAI’s most important safety principles- a prohibition on domestic mass surveillance and the requirement of human responsibility for the use of force, including autonomous weapon systems, are reflected in the agreement.

 

 

According to him, the Department of War agrees with these principles and incorporates them in law and policy. “We put them into our agreement,” he said, adding that the company will build technical safeguards to ensure its models behave as intended.

 

OpenAI will deploy field deployment engineers (FDEs) to support the models and ensure their safe use. The systems will operate on cloud networks only, with additional oversight mechanisms designed to prevent misuse.

 

Call for industry-wide standards

 

OpenAI has asked the Department of War to extend similar terms to all AI companies. In Sam Altman’s view, these are conditions that every firm in the sector should be willing to accept. He also expressed a preference for de-escalation, moving away from legal and governmental confrontation towards reasonable agreements.

 

Altman reiterated that OpenAI remains committed to serving humanity, while acknowledging that the world is “complicated, messy and sometimes dangerous”.

 

The agreement also comes against the backdrop of President Donald Trump’s recent decision to bar federal agencies from using Anthropic’s technology over national security concerns. Trump’s move came following differences over safeguards and operational restrictions on AI systems. Against this setting, OpenAI’s emphasis on prohibitions on domestic mass surveillance and human oversight of force positions its Pentagon agreement as an attempt to align commercial AI deployment with government requirements.