Former OpenAI safety researcher David Robinson has resigned and published an editorial arguing that the AI industry's culture is "broken," calling for nuclear-power-plant-level safeguards at frontier AI labs.
A former OpenAI safety employee has resigned and is publicly warning that the culture driving artificial intelligence development is fundamentally flawed, adding his voice to a growing chorus of researchers departing prominent AI firms.
David Robinson, who wrote the safety reports accompanying every major model release at OpenAI, announced his resignation this week and outlined his concerns in an editorial published in The Atlantic.
Robinson argued that Silicon Valley has operated with "extreme confidence" and "perpetual sprints," building larger and more capable models with "unimpeded optimism" that ignores or underestimates potential risks.
He called for the industry to develop humility and look beyond its insular, move-fast-and-break-things ethos — and drew a pointed comparison to industries with long records of managing catastrophic risk.
"Given today's risks, frontier labs need to run like nuclear power plants or busy airports, with layers of redundancy and careful, time-consuming planning, so that the occasional and inevitable human error does not open a door to disaster," Robinson wrote.
His departure follows a pattern that has accelerated over the past several months. Jacob Coxon appears to have initiated the recent wave of departures by quitting Anthropic and then publicly stating that AI "could kill us all by the end of the decade."
Robert O'Callahan, Bilal Chughtai, and Josh Engels subsequently left Google DeepMind. Joe Benton also departed Anthropic. Robinson's exit from OpenAI extends that sequence to one of the industry's most closely watched organizations.
OpenAI did not issue a public statement in response to Robinson's editorial, and the company has not publicly addressed his specific claims about its internal culture.
The timing of Robinson's resignation is notable. It comes as frontier AI labs are under intensifying regulatory scrutiny — the FTC has reportedly opened an investigation into both OpenAI and Anthropic — and as researchers across the field are publishing warnings about the risks posed by AI systems capable of self-improvement.
Robinson's core argument is that the problem is structural, not correctable by incremental policy adjustments. He contends the current moment calls not for new rules layered onto an existing culture, but for a wholesale rethinking of how AI development is governed and paced.
Whether that argument gains traction inside the organizations still building frontier models — or lands primarily with regulators and outside observers — will be a key indicator of how seriously the industry is prepared to treat the concerns of its own departing experts.
Sara Montes de Oca is the Editor in Chief of TechEchelon. Previously a correspondent and producer in Washington, D.C., covering business, finance, and politics.
More from Sara →Tesla's Cybercab fleet in Texas grew from 45 to 169 authorized vehicles in its first month of commercial operations in Austin, but mixed passenger reviews, federal safety scrutiny, and a dominant rival in Waymo underscore the challenges ahead for the driverless ride-hail service.
Public opposition to data centers is spreading from the U.S. to Europe and Asia, putting roughly $42 billion in European investments at risk through delays, cancellations, and tightening regulation, even as AI-driven demand continues to grow.
Google unveiled Gemini 4 Argon this week, a flagship AI model that analysts say returns the company to the frontier of large language model development — though not to the top of it — as competition with OpenAI and Anthropic intensifies.
Written for readers who already know the basics — markets, AI, and the policy decisions that shape both.