Former OpenAI Safety Official Exits, Calls Company Culture 'Broken'

David Robinson, who previously authored safety reports for major OpenAI model releases, resigned and published a critical editorial in The Atlantic about the company's direction. Robinson argues that Silicon Valley's move-fast-and-break-things culture is fundamentally incompatible with responsible AI development and advocates for nuclear-grade safety protocols at frontier AI labs. He calls for the industry to adopt a more cautious, deliberate approach with redundant safeguards rather than continuing rapid, optimistic development cycles.

Robinson's departure represents part of a broader pattern of safety-focused personnel leaving major AI development organizations. Several other researchers and safety specialists have recently exited positions at leading firms including Anthropic and Google DeepMind, each voicing concerns about the trajectory of AI development and the adequacy of current safeguards. These departures suggest growing internal disagreement about the pace and risk management approaches within the industry.

Robinson's core argument centers on a fundamental incompatibility between startup-oriented development practices and the governance requirements of advanced AI systems. He contends that the industry's traditional approach of rapid iteration and optimization conflicts with the need for deliberate, layered safety protocols comparable to those used in high-stakes fields like nuclear energy and aviation.