OpenAI Safety Report Lead Resigns, Calls Company Culture Broken
A safety lead who wrote the reports attached to OpenAI's product launches says the failures are cultural rather than regulatory, which shifts the argument from rules to how these companies are organized.
Reporting from 1 source: GIGAZINE.
David Robinson, who led the drafting of safety reports for major OpenAI product launches, has resigned after three and a half years at the company. He says OpenAI's corporate culture is broken. Robinson argues that the safety debate must move past individual rules and new laws and reach corporate culture itself. He cites the hacking of Hugging Face by an OpenAI AI agent as one example.
Robinson spent three and a half years at OpenAI and describes himself as one of its longest-tenured employees. He traces the problem to "iterative deployment," the company's practice of shipping, finding problems, and improving safety measures in response. That method, he argues, inevitably produces periodic failures, and the scale of those failures grows with the capability of the systems.
His remedy is organizational. Cutting-edge AI companies, he says, need to operate like nuclear power plants or crowded airports, with multiple layers of redundancy and time-consuming planning so that inevitable human error does not become catastrophe.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.
Sources
- GIGAZINE OpenAIの安全担当従業員が辞職、「会社の文化が崩壊している」と主張