Pentagon GenAI Mil

The Pentagon gave 3 million military and civilian personnel access to ChatGPT and Grok this week. The more interesting headline is who said no.

GenAI.mil launched in December with only Gemini running the platform. As of this week it added OpenAI's ChatGPT Mil and xAI's Grok for Government, and the department says 1.7 million people are already active users out of a 3 million-person workforce.

Anthropic was reportedly at the same table and walked. The company wanted contract language ruling out use of its models for mass surveillance of Americans or lethal autonomous weapons, terms the Pentagon wouldn't accept.

This is the first public case of a frontier AI lab treating "we won't sign this" as a governance decision rather than a talking point, on a contract worth real revenue and real legitimacy. For anyone watching how AI vendors will handle high-stakes deals in healthcare, finance, or critical infrastructure, this is a preview. The labs setting non-negotiable use restrictions now are defining what responsible AI actually costs them in market share.

Should AI companies be willing to walk away from lucrative government contracts over use restrictions, or is that a luxury only a handful of well-funded labs can afford?