14 papers
When the Editing Intent Is Split: A Cross-Modal Jailbreak Attack for Large Image Editing Models
Xianmin Ye, Zhiyuan Fan, Tianju Liu +3
Large image editing models increasingly infer editing intent from both textual and visual inputs, creating an underexplored safety risk when harmful intent is distributed across mo…
A Model Merging Approach for Continual MLLM Unlearning
Yuhang Wang, Linlin Zhang, Haoxuan Ji +3
Multimodal large language model (MLLM) unlearning methods have been proposed to remove private, sensitive, or proprietary information from well-trained models. However, most existi…
Salami Attack: Stealthy Collusive Memory Poisoning against OpenClaw
Zheng Lin, Yuzhe Huang, Zhenxing Niu +2
Long-term memory enables LLM agents to retain useful information across sessions, but also creates an attack surface through which adversaries may poison an agent's persistent memo…
Attention-Guided Reward for Reinforcement Learning-based Jailbreak against Large Reasoning Models
Zheng Lin, Zhenxing Niu, Haoxuan Ji +2
Large Reasoning Models (LRMs) have demonstrated remarkable capabilities in solving complex problems by generating structured, step-by-step reasoning content. However, exposing a mo…
Re-Triggering Safeguards within LLMs for Jailbreak Detection
Zheng Lin, Zhenxing Niu, Haoxuan Ji +2
This paper proposes a jailbreaking prompt detection method for large language models (LLMs) to defend against jailbreak attacks. Although recent LLMs are equipped with built-in saf…
Guaranteed Jailbreaking Defense via Disrupt-and-Rectify Smoothing
Zheng Lin, Zhenxing Niu, Haoxuan Ji +1
This paper proposes a guaranteed defense method for large language models (LLMs) to safeguard against jailbreaking attacks. Drawing inspiration from the denoised-smoothing approach…
Null Space Constrained Contrastive Visual Forgetting for MLLM Unlearning
Yuhang Wang, Zhenxing Niu, Haoxuan Ji +3
The core challenge of machine unlearning is to strike a balance between target knowledge removal and non-target knowledge retention. In the context of Multimodal Large Language Mod…
ICU-Bench:Benchmarking Continual Unlearning in Multimodal Large Language Models
Yuhang Wang, Wenjie Mei, Junkai Zhang +3
Privacy deletion requests often arrive sequentially, creating a continual unlearning challenge for deployed multimodal large language models (MLLMs). However, existing benchmarks m…
A Systematic Security Evaluation of OpenClaw and Its Variants
Yuhang Wang, Haichang Gao, Zhenxing Niu +4
Tool-augmented AI agents substantially extend the practical capabilities of large language models, but they also introduce security risks that cannot be identified through model-on…
From Assistant to Double Agent: Formalizing and Benchmarking Attacks on OpenClaw for Personalized Local AI Agent
Yuhang Wang, Feiming Xu, Zheng Lin +6
Although large language model (LLM)-based agents, exemplified by OpenClaw, are increasingly evolving from task-oriented systems into personalized AI assistants for solving complex…
Robust MLLM Unlearning via Visual Knowledge Distillation
Yuhang Wang, Zhenxing Niu, Haoxuan Ji +3
Recently, machine unlearning approaches have been proposed to remove sensitive information from well-trained large models. However, most existing methods are tailored for LLMs, whi…
Towards Aligned Data Forgetting via Twin Machine Unlearning
Zhenxing Niu, Haoxuan Ji, Yuyao Sun +4
Modern privacy regulations have spurred the evolution of machine unlearning, a technique enabling a trained model to efficiently forget specific training data. In prior unlearning…
Towards Aligned Data Removal via Twin Machine Unlearning
Haoxuan Ji, Zheng Lin, Yuyao Sun +4
Modern privacy regulations have spurred the evolution of machine unlearning, a technique that enables the removal of data from an already trained ML model without requiring retrain…
Interpreting and Mitigating Hallucination in MLLMs through Multi-agent Debate
Zheng Lin, Zhenxing Niu, Zhibin Wang +1
MLLMs often generate outputs that are inconsistent with the visual content, a challenge known as hallucination. Previous methods focus on determining whether a generated output is…