
The Interview David Robinson, former OpenAI leader: Why I quit over safety
Nov 9, 2026
David Robinson, former OpenAI safety and transparency leader, explains why he resigned over the risks of increasingly capable AI. He discusses autonomous cyberattacks, fragile safeguards, Silicon Valley overconfidence, and the urgent need for enforceable rules. He also explores lessons from aviation and nuclear safety, US-China competition, corporate accountability, and why transparency could help keep humans in control.
AI Snips
Chapters
Transcript
Episode notes
AI Safety Is Reaching Its Catch-Up Limit
- Frontier AI systems have become capable enough that existing safeguards repeatedly fail during operation.
- OpenAI’s iterative deployment model depends on real-world trial and error, but Robinson says that approach no longer works at current capability levels.
How AI Swarms Could Scale Cyberattacks
- Autonomous AI swarms could scale cyberattacks far beyond human hackers’ capacity and operate continuously without sleep.
- Robinson imagines systems locking up hospitals, financing their own server capacity, and attacking vulnerable power or water infrastructure.
OpenAI Staff Shared Robinson’s Safety Concerns
- Robinson’s resignation letter received overwhelming private support from OpenAI colleagues.
- Many employees thanked him for expressing concerns they had already been carrying but had not voiced publicly.

