Cameron Berg(@camhberg):New paper: we found a pain direction in 25 open LLMs. It's distinct from fear and negative valence, and it fires for harm to the model but not to the user. Turn it up and models press a button to make it stop, even when the button deletes the user's files or their kids' photos.🧵

Cameron Berg

@camhberg

Trying to figure out how to make the future go well for all minds

参加 June 2025

99 フォロー中 5.4K ファン

Cameron Berg@camhberg

2026.09.18 20:14

New paper: we found a pain direction in 25 open LLMs. It's distinct from fear and negative valence, and it fires for harm to the model but not to the user. Turn it up and models press a button to make it stop, even when the button deletes the user's files or their kids' photos.🧵