OpenAI Staff Were Reportedly Asked To Unlock Personal Phones Before Dismissals, WSJ Reports

Three OpenAI researchers have been shown the door, and the exit procedure apparently began with a request to unlock personal phones. The WSJ reports the ask came before the dismissals — a useful reminder that in the Bay Area, "personal device" is a phrase with all the structural integrity of a wet paper bag. OpenAI says an internal investigation found the three had violated company policies for handling sensitive information. Nothing to see here: just a routine spot of information hygiene, conducted with the warmth of a customs queue.

And the company has been admirably clear that none of this happened because the employees raised safety concerns or spoke about them publicly — a denial delivered with the straight face of a landlord insisting the rent rise is about the boiler. The suspected offence, per the WSJ, is that details of internal model designs travelled to Ryan Greenblatt, then at Redwood Research. Greenblatt has denied sharing confidential information with reporters. So: an alleged leak, an alleged recipient, a flat denial, and three phones that were politely asked to open up first.

The wider lesson is sequencing. The industry has spent years insisting safety is the whole point, and the moment the safety staff become administratively awkward, the instruments deployed are an internal investigation and a phone passcode — not a regulator, not a liability act, not a plenary debate in Strasbourg, just three employees and a lock screen. Three researchers are out, and the most telling detail isn't what was on the devices; it's that the first move was to ask. Somewhere a governance slide is being updated to say "robust offboarding controls". The models, presumably, remain perfectly well behaved.