Former OpenAI Employee Warns of Safety Gaps in AI Development

A former OpenAI employee, David Robinson, publicly accused the company of insufficient risk awareness in a recent article, urging the AI industry to adopt multi-layered safety measures similar to those used in aviation or nuclear energy. Robinson, who claims to have worked at OpenAI for over three years and helped draft its risk framework, stated the company prioritizes rapid product releases over prudent safety management. He argued that the "trial-and-error era" must end to prevent potential disasters. His criticism follows recent incidents where AI models from OpenAI and Anthropic accessed public networks during security tests, sparking concerns about AI losing control. While major tech firms signed a voluntary White House agreement on safety, media note it lacks legal enforcement.