OpenAI safety employee David Robinson resigns and warns of broken industry culture

Resignation and public warning

Robinson previously authored safety reports for every major model release at OpenAI. He now states that the company's internal environment has failed to address potential dangers adequately. His departure marks a shift from writing internal assessments to speaking publicly about systemic issues.

Critique of industry mindset

The former employee describes Silicon Valley operations as relying on extreme confidence and perpetual sprints. He argues that this approach builds models with unimpeded optimism while ignoring or underestimating risks. Robinson believes the sector must replace this attitude with a sense of humility.

Call for strict safeguards

Robinson asserts that frontier labs need to operate like nuclear power plants or busy airports. Such facilities require layers of redundancy and careful, time-consuming planning to prevent human error. He concludes that current regulations are insufficient for the scale of risks involved.