Revealing a Hidden Class of Task-in-Prompt Adversarial Attacks on LLMs(aclanthology.org)
2 pointsby aghuang2 hours ago1 comment