Xiaodong Li

The paper proposes DAMPER, a domain-aware framework that autonomously extracts and rewrites private information from text while providing rigorous differential privacy guarantees, significantly improving the privacy-utility trade-off.

The paper introduces a framework using the 'behavioral geometry' of model populations to efficiently predict jailbreak susceptibility and transfer defenses, achieving high accuracy with significantly fewer evaluations.

A cloud-scale gateway system for MCP services is presented, which breaks the direct-connect model and offloads legacy service integration, consolidates incompatible MCP variants, and reduces tool selection time and token usage.