David Robinson, the senior safety lead who drafted OpenAI’s safety assessments for every major product launch, has resigned, warning that the firm’s internal culture is “broken”.

Resignation and warning

Robinson announced his departure in an essay titled “I quit OpenAI because its culture is broken”, saying the company was not “being nearly careful enough” about the rapid rollout of increasingly powerful models.

“I quit OpenAI because its culture is broken”.

In the piece, he said that the pace of development often left little room for thorough risk analysis, and that internal pressures to ship new features outweighed caution.

Industry reaction and regulatory backdrop

Tech commentators have noted that Robinson’s exit follows a pattern of senior staff issuing public cautions when they leave AI firms. TechCrunch described him as “something of a cliché” – a senior employee sounding an alarm as they walk out the door.

Analysts say the resignation adds weight to calls from regulators in the US and Europe for tighter oversight of AI systems. The European Commission, which is drafting a comprehensive AI Act, has highlighted the need for robust internal safety processes at leading developers.

Implications for AI safety and users

Robinson’s departure could affect the timing of upcoming model releases, as his team was responsible for the safety reports that accompany each launch. Without his oversight, OpenAI may need to restructure its safety workflow, potentially slowing the introduction of new features.

For users, the warning underscores the risk that AI products could be deployed before all foreseeable harms are mitigated. Consumer‑facing applications such as ChatGPT and the new Whisper‑2 voice model may face heightened scrutiny from both the public and regulators.OpenAI has not yet commented on how it will address the concerns raised. The company’s board is expected to meet later this week to review internal governance, according to sources familiar with the matter.

The episode arrives at a time when the AI sector is grappling with high‑profile mishaps, from biased outputs to unintended content generation. Experts argue that a culture that prioritises speed over safety could stall broader adoption and invite stricter legislative action.