Yuguang Zhou

This paper introduces a novel framework, the Reasoning Safety Monitor, to detect and prevent logical inconsistencies and adversarial manipulations within the internal reasoning steps of large language models, establishing reasoning safety as a critical security dimension.

ZK-Value introduces a practical, scalable zero-knowledge system for calculating data valuations (Shapley values) in data marketplaces, significantly reducing proving time while maintaining high accuracy.

This paper reveals a denial-of-service vulnerability in LLM-based guardrails for autonomous agents and proposes two attack frameworks.

AutoSpec is a framework that uses counterexample-guided inductive synthesis and inductive logic programming to automatically evolve safety rules for large language model agents.