Yulin Li

The paper introduces CanaryRAG, a novel dual-path runtime defense mechanism that detects RAG Knowledge Base Leakage attacks by embedding canary tokens into retrieved knowledge chunks.

The paper introduces MechVQA, a comprehensive dataset and benchmark for mechanical drawing understanding, and proposes the MechVL model, which significantly improves Multimodal LLMs' performance on these specialized tasks.

The paper introduces and analyzes cross-session stored prompt injection, demonstrating that persistent system state transforms prompt injection from a temporary model-level threat into a long-lived, system-level vulnerability in agentic systems.