Revealing a Hidden Class of Task-in-Prompt Adversarial Attacks on LLMs

2 pointsposted 9 hours ago
by aghuang
(aclanthology.org)

1 Comments