Revealing a Hidden Class of Task-in-Prompt Adversarial Attacks on LLMs(aclanthology.org)2 points by aghuang 13 days ago | 0 comments