OpenAI, the maker of ChatGPT, has announced an agreement with the U.S. Department of War to deploy its AI models within the department’s classified network.
According to OpenAI, the deal includes safety guardrails such as prohibitions on domestic mass surveillance and a requirement for human responsibility in the use of force, including in matters involving autonomous systems. The company said these principles were reflected in the final agreement and that deployment will occur within secure, cloud-based environments with additional technical safeguards.
However, the development is drawing heightened attention because the structure of the agreement mirrors the same conditions that reportedly led to the Department of War designating rival AI firm Anthropic a “supply chain risk.” Anthropic had previously resisted broader federal access to its models, citing ethical restrictions around surveillance and weaponization, a stance that triggered federal escalation and restrictions.
Despite that earlier crackdown, the Department has now accepted similar guardrail-based terms from OpenAI, effectively approving an arrangement that aligns with the safety principles Anthropic publicly advocated.
The contrast is likely to fuel debate over consistency in federal AI policy, competitive dynamics within the AI industry, and the evolving standards governing military use of advanced artificial intelligence.
The agreement signals a major shift in the relationship between U.S. defense institutions and frontier AI companies, and could reshape how AI governance, national security, and corporate policy intersect going forward.
