OpenAI Built Robot Brains Then Got Weird About Who Uses Them

A senior OpenAI robotics engineer reportedly quit because the “military guardrails” were basically vibes and a prayer—right after OpenAI started playing footsie with the U.S. defense world.

Nothing says “benefit humanity” like discovering your safety plan is just a Slack emoji and a Terms of Service PDF.

Here’s the alleged plot: OpenAI has been talking up responsible AI, then enters a defense partnership orbit, and suddenly one of the grown-ups in the robotics room decides they’re not staying to watch the “don’t worry, we’ll be careful” phase turn into the “oops, it shipped” phase.

And robotics isn’t just chatbots writing your emails. Robotics is AI that can touch the real world. It can move stuff, surveil stuff, block stuff, chase stuff. You know, the kind of stuff governments absolutely never misuse and definitely won’t point at protestors, borders, or anyone who looks “statistically suspicious.”

OpenAI’s whole brand is “we have rules.” Cool. The complaint here is those rules apparently weren’t specific enough for military deployment. Which is like saying the guardrails on a rollercoaster are “a little loose” right before the drop.

Meanwhile, regular people get told AI is coming for their jobs because “efficiency,” but the best-funded use case is still “help the state do state things, faster.”