The article reports that OpenAI has provided details on a newly signed agreement with the U.S. Department of Defense to deploy its artificial intelligence technology on the military’s classified networks. The announcement comes amid a tense dispute between the Pentagon and another AI lab, Anthropic, which was recently designated a “supply-chain risk” by the Trump administration after refusing certain terms on AI use.
OpenAI stated that its contract includes “layered protections” and three clear red lines to ensure responsible use of its AI in defence settings. Specifically, the deal prohibits the AI from being used for mass domestic surveillance, directing autonomous weapons systems, or making high-stakes automated decisions without human oversight. These safeguards are aimed at addressing ethical concerns while allowing the technology to support national security missions.
To enforce these safety measures, OpenAI says it will retain full control over its safety stack, deploy through secure cloud infrastructure, and keep cleared OpenAI personnel in the loop during use — all backed by strong contractual language. The company also warned that any breach of the contract by the U.S. government could trigger termination, though it expressed confidence that such a scenario is unlikely.
The broader context includes the Pentagon signing similar agreements — each worth up to $200 million — with multiple major AI labs as part of efforts to integrate advanced AI tools into defence workflows. OpenAI also publicly objected to the government’s designation of Anthropic as a supply-chain risk, saying the start-up should not be labeled as such.
