Longzhu He

Large language models (LLMs) can effectively handle outdated information through knowledge editing. However, current approaches face two key limitations: (I) Poor generalization: Most approaches rigidly inject new knowledge without ensuring that the model can use it effectively to solve practical...

Retrieval-Augmented Generation (RAG) systems are vulnerable to perspective manipulation attacks in multi-perspective long-form generation scenarios: adversaries inject documents with implicit stance biases to induce the system to generate content favoring specific perspectives. While existing RAG...