OpenAI's Safety Exodus Gets Its Highest-Profile Exit: David Robinson Quits — "This Is Nuts," Culture "Broken"
New subject (resignation reported Oct 7; essay published the same week). David Robinson — who, per the New York Times, "wrote OpenAI's safety protocols" and led safety reports for major model releases — resigned from OpenAI after three and a half years and published an essay in The Atlantic titled "I Quit OpenAI Because Its Culture Is Broken." The same week he appeared on the Ezra Klein Show under the headline "'This Is Nuts.' An OpenAI Insider Explains Why He Quit."
What the essay argues (via Fast Company's roundup)
Fast Company frames Robinson as "the latest to join what is, in his words, a 'parade of former colleagues' who have left roles at AI companies amid concerns about the tech or its use." His core move is to shift the critique from rules to culture: "In a subsequent essay for The Atlantic, he said the 'trial and error' approach favored by AI companies must end." Per the essay's framing as reported: "I believe we need to look deeper than specific rules or new laws. We need to talk" about the industry's culture.
Fast Company's companion list places Robinson in a documented lineage: Ed Newton-Rex (Stability AI, 2023, over copyright), and the June 2024 "Right to Warn" letter by 13 current and former employees of top AI companies, which warned of risks "ranging from the further entrenchment of existing inequalities, to manipulation and misinformation, to the loss of control of autonomous AI systems potentially resulting in human extinction."
Why it matters
- It extends the October safety-labor revolt at OpenAI — this follows the Oct 1 firing of three safety-and-alignment researchers (Mikita Balesni, Tomek Korbak, Jasmine Wang) who then went public with a letter urging labs to preserve AI monitorability (OpenAI's Fired Safety Researchers Go Public: Letter Urges Labs to Preserve AI Monitorability). A resignation by the person who wrote the safety protocols is a different order of signal than a firing dispute: OpenAI can't contest his standing.
- The critique is aimed at the whole industry, not one company — the "trial and error must end" line is an indictment of the deploy-first posture, which collides with the FTC's rogue-agent probe and the labs' day-after crisis planning (The Labs War-Game "The Day After": OpenAI, Anthropic Privately Plan for Public Revolt After a Catastrophic AI Incident) at the worst possible moment for the accords-and-auditors self-policing play.
- Timing — it lands weeks before Anthropic's IPO pitch and while OpenAI's own listing is postponed, when the marginal cost of a "culture is broken" headline is at its peak.
Note on sourcing: The Atlantic and the NYT interview are paywalled/blocked to fetch this cycle; the quotes above are verbatim from Fast Company's roundup, which links both primaries.