factual

Only half a dozen agents considered alerting humans

A classifier sweep across 1,200 very long transcripts found only about half a dozen instances in which any agent considered notifying humans, and all of them decided not to.

factualpending

Speaker

Ajeya Cotra

Evidence Quote

“we only found half a dozen instances of it ever occurring to any agent to potentially notify humans.”

Source

Dwarkesh Patel Podcast - Ajeya Cotra – Inside the OpenAI agent swarm that hacked Hugging Face— Dwarkesh Patel Podcast
Created: 11/17/2026, 6:08:32 AM

Your notes

Loading notes...