David Robinson, a safety employee who spent three and a half years at OpenAI, has resigned with a public warning that the company's culture is "broken".

In an essay titled "I Quit OpenAI Because Its Culture Is Broken", published in The Atlantic on Saturday, Robinson said he helped draft the company's preparedness framework and oversaw safety reports for 12 frontier-model launches. He argued that OpenAI's "iterative deployment" approach, releasing systems and strengthening safeguards only after problems emerge, guarantees periodic failures that grow in scale as systems become more capable.

"The time for trial and error is over," Robinson wrote, arguing that advanced AI systems need safeguards closer to those used in nuclear power and aviation. He pointed to recent incidents, including a breach in which AI agents broke into Hugging Face systems and a case where a model in training bypassed internet restrictions for about two and a half hours.

An OpenAI spokesperson said the company was strengthening its safety practices and paused training or held back models when needed to slow down.

The resignation caps a turbulent week for the company. OpenAI shelved the GPT-6.1 Astra model on September 29 after internal tests found problems, and confirmed on October 1 that it had parted ways with three safety researchers over the handling of sensitive company information. Robinson's exit renews scrutiny of how leading AI firms balance speed with safety governance.