OpenAI readies ChatGPT deployment for 3 million Pentagon personnel

The deployment represents one of the largest single-organization rollouts of a commercial LLM to date. OpenAI's custom ChatGPT will go live on GenAI.mil — the Pentagon's in-house generative-AI platform — in early July, reaching more than 3 million defense personnel. The build is certified for unclassified work and runs entirely within authorized government cloud infrastructure, a requirement for handling sensitive but unclassified workflows.

The initiative is a joint effort with the DoD's Chief Digital and Artificial Intelligence Office (CDAO), which has been standardizing how the military procures and deploys AI. It complements the earlier availability of OpenAI's models — GPT-5.4, GPT-5.5, and Codex — to the federal workforce via Amazon Bedrock and GovCloud, giving the government multiple procurement paths for OpenAI capabilities.

Competitively, the deal cements OpenAI's lead in U.S. government adoption at a moment when rivals face headwinds: Anthropic's models were just pulled by export-control order, and xAI's Grok is mired in controversy over its targeting role. The contrast is stark — OpenAI is scaling a sanctioned, unclassified productivity deployment while competitors grapple with national-security entanglements. The key questions to watch: what guardrails govern 3 million users on a single platform, how usage is audited, and whether the unclassified scope holds or expands toward more sensitive work over time.