1 paper
Zhifeng Jiang, Zhihua Jin, Guoliang He
System prompts are widely used to guide the outputs of large language models (LLMs). These prompts often contain business logic and sensitive information, making their protection e…