OpenAI Went Classified With the Pentagon and One Robot Guy Rage-quit

OpenAI just signed a classified deployment deal with the U.S. Department of Defense, and a senior robotics team member quit like he’d just watched a Roomba get handed a drone controller.

Yes, the “AI safety” shop that swears it’s building guardrails just walked into the military black box where guardrails go to get vaporized by paperwork.

Because the agreement is classified, we get the usual vague adult-language: “deployment,” “partnership,” “national security,” the whole bedtime story for taxpayers.

The resigning staffer’s complaint wasn’t “I hate defense.” It was “the safety and governance isn’t there.” Which is honestly worse, because that’s not ideology—that’s an engineer saying, “this thing can hurt people and you’re asking me to ship it anyway.”

Meanwhile, this is exactly how you end up with an AI company whose public brand is “trust us” and whose biggest customer is the one institution on Earth legally allowed to say “trust us” while doing the most untrustworthy stuff imaginable.

Also: if a senior robotics person is walking out, imagine what the rank-and-file feels like when the moral compass points straight to ‘C’ for Classified and HR hands you a mindfulness worksheet.

And yes, this absolutely lands on your life. The same models that get “deployed” behind closed doors end up in the open market, except now the incentives are military-grade secrecy and contract money, not your rights, your job, or whether you want your future governed by a Terms of Service written like a hostage note.