OpenAI strikes classified AI deal with US Department of War, cites safety guardrails

OpenAI

• OpenAI to deploy AI models in US Department of War's network
• Deal bans mass surveillance and autonomous weapons
• OpenAI calls for uniform contract terms for AI firms in defense deals

Did our AI summary help?

OpenAI has announced that it has reached an agreement with the US Department of War to deploy its AI models inside the department’s classified network, a move that significantly deepens the company’s involvement in national security infrastructure.

In a post on X, OpenAI CEO Sam Altman said: “Tonight, we reached an agreement with the Department of War to deploy our models in their classified network.” The company described the negotiations as constructive, adding that the DoW showed “a deep respect for safety and a desire to partner to achieve the best possible outcome.”

Guardrails written into the agreement

Altman stressed that its core safety principles remain intact. Two in particular were highlighted: prohibitions on domestic mass surveillance and the requirement for human responsibility in the use of force, including autonomous weapon systems.

“The DoW agrees with these principles, reflects them in law and policy, and we put them into our agreement,” Altman said.

The deployment will include technical safeguards to ensure model behaviour aligns with these commitments. OpenAI also confirmed it will deploy field deployment engineers to support usage and monitor safety, and that the models will run on cloud networks only.

In a notable move, OpenAI said it is asking the Department of War to offer the same contractual terms to all AI companies. “We have expressed our strong desire to see things de-escalate away from legal and governmental actions and towards reasonable agreements,” the company stated, an apparent reference to ongoing tensions between AI firms and regulators.

The announcement inevitably draws comparisons with Anthropic, which has previously emphasised strict usage policies around military and intelligence applications. Anthropic has positioned itself as cautious about defence deployments, focusing heavily on constitutional AI and structured safeguards.

OpenAI’s agreement suggests a more direct engagement with government agencies, albeit with publicly articulated red lines. The competitive dynamic between major AI labs is now extending beyond consumer products and enterprise contracts into classified and defence domains.

Trump’s AI and defence stance

The deal also lands in a political context shaped by President Donald Trump, who has repeatedly argued that the United States must lead in artificial intelligence for national security reasons. Trump has framed AI dominance as essential to maintaining military and economic superiority, while warning against overregulation that could slow American innovation.

Altman concluded his statement with a broader reflection: “We remain committed to serve all of humanity as best we can. The world is a complicated, messy, and sometimes dangerous place.”

That tension sits at the heart of this agreement. AI companies are increasingly central to geopolitical competition, yet they continue to frame their missions around global benefit and safety.

The question now is not whether advanced AI will play a role in defence infrastructure. It already is. The debate is about the guardrails, the oversight and whether these principles hold when theory meets real-world pressure.