9 papers
Turning Bias into Bugs: Bandit-Guided Style Manipulation Attacks on LLM Judges
Xianglin Yang, Bryan Hooi, Gelei Deng +2
The known stylistic biases in LLM judges, such as a preference for verbosity or specific sentence structures, present an underexplored security vulnerability. In this work, we intr…
Inverting the Shield: Systematically Generating Safety Tests from Policy Specifications
Xiaoyue Lu, Xianglin Yang, Haijun Liu +4
The widespread integration of Large Language Models (LLMs) necessitates rigorous and systematic safety evaluation. Existing paradigms either rely on constructed benchmarks to asses…
Zombie Agents: Persistent Control of Self-Evolving LLM Agents via Self-Reinforcing Injections
Xianglin Yang, Yufei He, Shuo Ji +2
Self-evolving LLM agents update their internal state across sessions, often by writing and reusing long-term memory. This design improves performance on long-horizon tasks but crea…
LLM-enabled Applications Require System-Level Threat Monitoring
Yedi Zhang, Haoyu Wang, Xianglin Yang +2
LLM-enabled applications are rapidly reshaping the software ecosystem by using large language models as core reasoning components for complex task execution. This paradigm shift, h…
Sparse Layer Sharpness-Aware Minimization for Efficient Fine-Tuning
Yifei Cheng, Xianglin Yang, Guoxia Wang +5
Sharpness-aware minimization (SAM) seeks the minima with a flat loss landscape to improve the generalization performance in machine learning tasks, including fine-tuning. However,…
TraceAegis: Securing LLM-Based Agents via Hierarchical and Behavioral Anomaly Detection
Jiahao Liu, Bonan Ruan, Xianglin Yang +5
LLM-based agents have demonstrated promising adaptability in real-world applications. However, these agents remain vulnerable to a wide range of attacks, such as tool poisoning and…
AutoContext: Instance-Level Context Learning for LLM Agents
Kuntai Cai, Juncheng Liu, Xianglin Yang +3
Current LLM agents typically lack instance-level context, which comprises concrete facts such as environment structure, system configurations, and local mechanics. Consequently, ex…
Enhancing Model Defense Against Jailbreaks with Proactive Safety Reasoning
Xianglin Yang, Gelei Deng, Jieming Shi +2
Large language models (LLMs) are vital for a wide range of applications yet remain susceptible to jailbreak threats, which could lead to the generation of inappropriate responses.…
Neural Surveillance: Live-Update Visualization of Latent Training Dynamics
Xianglin Yang, Jin Song Dong
Monitoring the inner state of deep neural networks is essential for auditing the learning process and enabling timely interventions. While conventional metrics like validation loss…