September 1, 2026ToolInfrastructure

The Military's ChatGPT Is Live. Claude Isn't Invited

On August 31 the Pentagon flipped the switch: ChatGPT Mil and Grok for Government are now live on GenAI.mil, the Defense Department's platform for giving its people frontier models without routing sensitive data through consumer channels. The scale is not a pilot: built for 3 million civilian and military personnel, and the platform already counts 1.7 million unique users from its Gemini-only first phase. ChatGPT Mil is scoped to unclassified document-heavy work: planning, policy, logistics, administration.

The name is the story. OpenAI didn't sell the Pentagon API access, it shipped a "Mil" variant tailored, in the DoD's words, to warfighter needs. xAI did the same with Grok. That's a product tier that didn't exist a year ago: frontier labs building military editions as first-class SKUs.

And one lab is conspicuously absent. No Claude on GenAI.mil, two days after a federal judge ruled the Pentagon's blacklisting of Anthropic illegal (https://clauday.com/article/cf7e9a57-71ee-4421-a0df-34153d6678e0). The split is now fully legible at the product level: OpenAI and xAI ship Mil variants, Anthropic insists on guardrails and meets the Pentagon in court. Three million users just became the measure of what that principle costs, or what it's worth.

TechCrunch coverage: https://techcrunch.com/2026/08/31/the-pentagon-now-has-its-own-version-of-chatgpt-and-grok/
← Previous
Simon Willison Made ChatGPT Work Publish Its Own Manual
Next β†’
OpenAI Is Buying Mac Minis by the Tens of Thousands
← Back to all articles

Comments

Loading...
>_