Former OpenAI safety staffer David Robinson resigns and criticizes safety culture

In short: David Robinson resigned from OpenAI and publicly criticized how the company handles safety as it builds more powerful AI.

What happened

Reporting published on October 3, 2026, said David Robinson left OpenAI the previous week. In an essay for The Atlantic titled “I Quit OpenAI Because Its Culture Is Broken,” he argued that OpenAI is moving too quickly and relies too much on trial and error.

Robinson said he led the writing of safety reports that were released alongside major OpenAI product launches. These reports are meant to explain risks and what the company did to reduce them, like a safety label that comes with a new appliance.

Some posts and descriptions have claimed Robinson “wrote OpenAI’s safety protocols.” Based on the available reporting, that wording goes further than the facts support. The reports indicate he worked on and led launch-related safety reports, but they do not establish that he authored the company’s safety rules overall.

OpenAI responded that it is strengthening safety and security practices. The company said it is expanding third-party evaluations (outside groups checking its work), increasing monitoring, and will pause training or hold back models when needed.

Why it matters

Many people now use AI tools at work, in school, and in everyday apps. When insiders disagree about whether safety work has enough time and authority, it can affect how much trust the public should place in these systems, and how carefully companies test them before release.

Source: NYTimes