1 paper
Zhuowei Chen, Qiannan Zhang, Shichao Pei
Jailbreak backdoor attacks on LLMs have garnered attention for their effectiveness and stealth. However, existing methods rely on the crafting of poisoned datasets and the time-con…