OpenAI to deploy AI models on classified U.S. military networks

OpenAI has reached a deal to deploy its AI models on U.S. Department of War classified cloud networks, marking a major escalation in military AI integration and a striking contrast to Anthropic's recent clash with the Pentagon. The agreement includes explicit contractual red lines: no autonomous weapons, no domestic mass surveillance, and no autonomous critical decision-making. Sam Altman praised the Department's "respect for safety and its cooperative approach during the agreement process."

The deployment matters because it signals how geopolitical competition is reshaping AI governance. Until now, concerns about military AI use centered on which company gets access. This deal suggests the real leverage lies with companies that cooperate on government terms. OpenAI retains control over its "safety stack"—the layered technical, policy, and human controls governing model behavior—and will determine which models are deployed and how safeguards are implemented. The deployment is limited to cloud environments, not edge systems like drones or aircraft. This comes weeks after the Trump administration designated Anthropic a "supply-chain risk to national security" and ordered federal agencies to cease using its technology within six months.