“AIs should also refuse to work on capabilities research” by Davidmanheim

/ LessWrong (Curated & Popular)