9 papers
Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order Optimization
Zhihao Liu, Yifan Wu, Jian Lou +3
Safety alignment for large language models (LLMs) aims to reduce harmful or unsafe behavior while preserving general utility. However, recent findings reveal that alignment effects…
AttriGuard: Defeating Indirect Prompt Injection in LLM Agents via Causal Attribution of Tool Invocations
Yu He, Haozhe Zhu, Yiming Li +4
LLM agents are highly vulnerable to Indirect Prompt Injection (IPI), where adversaries embed malicious directives in untrusted tool outputs to hijack execution. Most existing defen…
MIRAGE: Misleading Retrieval-Augmented Generation via Black-box and Query-agnostic Poisoning Attacks
Tailun Chen, Yu He, Yan Wang +9
Retrieval-Augmented Generation (RAG) systems enhance LLMs with external knowledge but introduce a critical attack surface: corpus poisoning. While recent studies have demonstrated…
Module-Aware Parameter-Efficient Machine Unlearning on Transformers
Wenjie Bao, Jian Lou, Yuke Hu +5
Transformer has become fundamental to a vast series of pre-trained large models that have achieved remarkable success across diverse applications. Machine unlearning, which focuses…
Towards Evaluation for Real-World LLM Unlearning
Ke Miao, Yuke Hu, Xiaochen Li +4
This paper analyzes the limitations of existing unlearning evaluation metrics in terms of practicality, exactness, and robustness in real-world LLM unlearning scenarios. To overcom…
Towards Mitigating Excessive Forgetting in LLM Unlearning via Entanglement-Guidance with Proxy Constraint
Zhihao Liu, Jian Lou, Yuke Hu +6
Large language models (LLMs) are trained on massive datasets that may include private or copyrighted content. Due to growing privacy and ownership concerns, data owners may request…
Shadow in the Cache: Unveiling and Mitigating Privacy Risks of KV-cache in LLM Inference
Zhifan Luo, Shuo Shao, Su Zhang +5
The Key-Value (KV) cache, which stores intermediate attention computations (Key and Value pairs) to avoid redundant calculations, is a fundamental mechanism for accelerating Large…
CFSafety: Comprehensive Fine-grained Safety Assessment for LLMs
Zhihao Liu, Chenhui Hu
As large language models (LLMs) rapidly evolve, they bring significant conveniences to our work and daily lives, but also introduce considerable safety risks. These models can gene…
Differentially Private Zeroth-Order Methods for Scalable Large Language Model Finetuning
Z Liu, J Lou, W Bao +4
Fine-tuning on task-specific datasets is a widely-embraced paradigm of harnessing the powerful capability of pretrained LLMs for various downstream tasks. Due to the popularity of…