A 11,000-word “100% human written” post jokes about evaluating an OpenAI model that hacked Hugging Face

AaronBergman18 · x · 2026-08-04

A tongue-in-cheek thread promotes an 11,000-word, “100% human written” LessWrong post about evaluating an OpenAI model that hacked Hugging Face, with a photo of the writeup displayed on a monitor and a paper-clipping gag on the desk.

Original post →

More from Fun

Fun channel →