Our agreement with the Department of War

Source: OpenAI Date: 2026-02-28 URL: https://openai.com/index/our-agreement-with-the-department-of-war

Summary

OpenAI’s February 2026 announcement of an agreement with the US Department of Defense (titled “Department of War” in the post, reflecting either a formal name change or OpenAI’s own framing choice). The agreement marked a significant policy shift from OpenAI’s earlier positions — prior usage policies had excluded military and weapons applications, and this agreement represented OpenAI’s formal entry into defense contracting. The arrangement likely covered cybersecurity assistance, logistics optimization, and intelligence analysis rather than kinetic weapons targeting.

Implications

The line OpenAI crossed. OpenAI’s defense agreement was one of the most consequential policy decisions in the company’s history, departing from a longstanding public position against direct military use. The timing — early 2026, in a changed political environment — reflected both commercial opportunity and a reassessment of what “beneficial for humanity” meant in a geopolitically contested AI landscape.

Thread: AI governance and dual-use policy. Sits alongside the Trusted Access for Cyber announcements, the Preparedness Framework’s national security provisions, the FedRAMP authorization, and the cybersecurity in the Intelligence Age piece as OpenAI’s systematic engagement with government and defense use cases.

Watch: What specific use cases were authorized under the DoD agreement and what ethical constraints OpenAI retained, whether the agreement included autonomous weapons systems or was explicitly limited to logistics/cyber/analysis, and how it affected OpenAI’s ability to retain safety-focused researchers.