Goal Hijacking Attack on Large Language Models via Pseudo-Conversation Injection.

Zheng Chen, Buhui Yao

Browse the full TrustCom paper archive.