David Robinson, the OpenAI employee responsible for writing safety reports on every major model release, has resigned and published a public indictment of the industry in The Atlantic.
Robinson's argument is not about missing regulations or insufficient guardrails. He says the culture is broken at a structural level, a claim that carries weight precisely because he was inside the process, not observing it from outside.
The cynic's reflex is to dismiss ex-employees who warn about what they helped build. Robinson's piece in The Atlantic is worth reading in full because the argument goes beyond the usual call for oversight and into the mechanics of how safety culture actually erodes from within.
[READ ORIGINAL →]