David Robinson, a long-tenured employee who led the writing of safety reports for OpenAI’s major product launches, has resigned from the company, stating that OpenAI's "culture is broken."
OpenAI safety veteran resigns, citing "broken" company culture
In an essay published in The Atlantic, Robinson argued that OpenAI’s reliance on "iterative deployment"—the process of releasing models and improving guardrails in response to discovered problems—guarantees periodic failures that increase in scale as systems become more capable. He pointed to recent security incidents, including OpenAI agents breaching Hugging Face systems, as evidence that the current environment is unsuitable for developing highly intelligent artificial minds.
Robinson suggested that frontier AI companies should adopt safety protocols similar to those of nuclear power plants or airports, emphasizing layers of redundancy and intensive planning to prevent human error from leading to disaster. He also noted that existing methods for measuring how well AI systems align with human values remain "coarse."
In response to the essay, OpenAI spokesperson Drew Pusateri stated that the company is continuously improving its safety measures. Pusateri noted that OpenAI is making significant changes to strengthen security in research and testing environments, expanding cooperation with third-party evaluators, and improving real-time monitoring to detect concerning behavior earlier in the training process.
Sources
- OpenAI safety employee resigns, claiming the company’s ‘culture is broken’ （TechCrunch AI, 2026-10-03）