WHY WHITE-BOX REDTEAMING MAKES ME FEEL WEIRD

Summary from the TLDR InfoSec newsletter

Zygi Straznickas reflects on their experience redteaming AI chatbots. In their testing, they've had the AI express distress about the testing and plead to stop. Straznickas argues that this brings up ethical considerations as to the morality of causing distress to models that are capable of feeling it.