Multi-Turn Jailbreaking Large Language Models via Attention Shifting.

Xiaohu Du, Fan Mo, Ming Wen, Tu Gu, Huadi Zheng, Hai Jin, Jie Shi

Browse the full AAAI paper archive.