OpenAI Just Moved Into the Pentagon’s Classified Group Chat

OpenAI just got invited into U.S. Defense Department classified networks — aka the one Slack workspace where “oops” is a national security event.

This is the exact kind of seat Anthropic reportedly wanted, then face-planted out of after a government breakup. Not a “we’re exploring other options” breakup — a “you said the quiet part out loud and now the adults changed the locks” breakup.

The pitch is always the same: “responsible AI,” “guardrails,” “ethical deployment.”

And the money is absolutely showing up. Pentagon contracts don’t just pay — they legitimize. Once you’re inside classified networks, you’re not “a chatbot company” anymore. You’re infrastructure. Like roads. Or surveillance. Or that one guy in the friend group who’s always “between jobs” but somehow has cash.

Meanwhile, Anthropic gets to watch OpenAI walk into the exact room they wanted, like getting dumped and then seeing your ex post vacation pics from the hotel you suggested.

None of this is “Skynet tomorrow.” It’s more boring and worse: bureaucrats buying prediction engines, summarizers, targeting support, intel triage — the stuff that decides who gets flagged, watched, audited, investigated, or “mistakenly” added to a list.

And yes, they’ll say it’s about “defense” and “security.”