Harmful Prompt Laundering: Jailbreaking LLMs with Abductive Styles and Symbolic Encoding.

Seongho Joo, Hyukhun Koh, Kyomin Jung

Browse the full EMNLP paper archive.