Our agreement with the Department of War

OpenAI has finalized an agreement with the Department of War (DoW) to deploy advanced AI systems in classified environments, insisting that the terms be made available to all AI companies. This agreement is guided by three core redlines: no use of OpenAI technology for mass domestic surveillance, autonomous weapons systems, or high-stakes automated decisions like social credit. Unlike other labs that rely primarily on usage policies, OpenAI's approach employs a multi-layered defense, retaining full discretion over its safety stack, deploying via the cloud, keeping cleared OpenAI personnel in the loop, and securing strong contractual protections. The deployment architecture is cloud-only, preventing use on edge devices for lethal autonomous weapons. The contract explicitly forbids using the AI system to independently direct autonomous weapons where human control is required by law, or to assume other high-stakes decisions needing human approval. Furthermore, intelligence activities must comply with existing laws like the Fourth Amendment, and the system cannot be used for unconstrained monitoring of U.S. persons' private information or domestic law enforcement beyond legal permissions. OpenAI believes this collaboration is necessary for the US military to counter adversaries integrating AI, while ensuring responsible deployment through robust safeguards.

(Source：OpenAI)