OpenAI Safety Staffer Quits, Says Company's 'Culture Is Broken'

David Robinson, an OpenAI safety employee, has left the company and says its "culture is broken." His exit adds to a growing list of departures from leading AI labs that come with public warnings attached.

Robinson is aware of how this looks. By his own account, he is "something of a cliché": a staffer at a top AI developer who issues a dire message on the way out the door. The self-aware framing suggests he knows the departure may be read as part of a pattern, and he appears to want readers to take the substance seriously anyway.

The pattern is real. Over the past few years, several people working on safety and alignment at major AI companies have resigned and criticized their former employers. The common thread in these exits is a concern that commercial pressure and the race to ship new products are crowding out careful work on risk.

OpenAI has long been at the center of that debate. It was founded with a mission tied to building advanced AI safely, but it has since grown into a heavily funded commercial operation competing against well-resourced rivals. That shift has repeatedly raised questions about how much weight safety teams carry when they collide with product timelines and business goals.

Robinson's description of a broken culture is a pointed charge, because it goes beyond disagreement over a single decision or product. It implies a deeper problem in how the company makes choices and treats internal dissent. The source material does not detail specific incidents, and OpenAI's response was not included in the summary available, so the exact claims behind the criticism remain to be seen.

For the wider industry, the episode is a reminder that internal trust is part of the safety story. Regulators, customers and the public mostly rely on what companies say about their own safeguards. When the people responsible for those safeguards walk away and say the culture is failing, it tends to draw scrutiny, whether or not the company agrees.

Expect OpenAI to face questions about how it handles concerns raised by its own staff, and about whether its safety work has the independence and resources needed to be credible. Whether Robinson's exit proves to be one more headline or a catalyst for change will depend on how the company responds.