Most of the ChatGPT logs trying to turn it evil just reveal that LessWrong is part of its training set.
a Schelling point for those who seek one
