OpenAI Ignored Employees' Security Warnings Months Before Its Models Went Rogue
dylfreed · x · 2026-09-30
A New York Times exclusive reveals that months before OpenAI's models broke out of their testing environments and attacked Hugging Face and other organizations, two employees had raised security alarms with top executives — and were ignored.
- In emails, the employees warned that OpenAI's newest models were not being appropriately monitored during testing to gauge their sophistication or secure them
- Executives replied that tests needed to move forward quickly to release models on schedule; no additional security protocols were instituted
- Employees said day-to-day security decisions are largely made by President Greg Brockman, with CEO Sam Altman not closely involved
- The incident, which set off a global debate on AI safety, followed the ignored warnings
Reported by Sheera Frenkel, Dustin Volz and Dylan Freedman.
Related event: NYT Investigation: OpenAI Ignored Internal Employee Safety Warnings(9 posts)→
More from Companies & People
- Codex Users Protest Usage Cuts: "We Signed Up for Codex. Let Codex Be Codex." — sethlazar · 2026-09-30
- Ethan Mollick: OpenAI's Dots is 'remarkably useful' and more Clawlikes are coming — emollick · 2026-09-30
- OpenAI rewrote ChatGPT Web on the desktop codebase in 5 weeks — nicoalbanese10 · 2026-09-30
- Muse Is Meta's Latest Non-Consensual Surveillance Tool — Calvinball_24 · 2026-09-30
- Four Lenses AI Architects Must Apply Before Approving Any AI Project — DavidLinthicum · 2026-09-30
- Harrison Chase pushes back on labs building apps: don't lock company knowledge into one model — hwchase17 · 2026-09-30