2 citations · 6 across the 19 of their papers we have counts for
23 papers
CoRedSkill: Stealthy Skill Composition Attacks via AST-Guided Injection
Qichen Zhao, Shengfang Zhai, Qingni Shen +2
Agents extend their capabilities through reusable skills that package instructions, executable code, and resources. Composing these skills enables complex task workflows, but their…
zkLLMPoT: Efficient Zero Knowledge Proof of Training for Large Language Models
Junkai Liang, Zhanpeng Guo, Pengfei Wu +5
Auditing the claimed outcomes of large language model (LLM) training is challenging when model weights and training data are private, while cryptographically proving the full train…
Membership Inference in Fine-tuned Diffusion Language Models via Token-level Memorization Asymmetry
Shengfang Zhai, Leo Marchyok, Yuling Shi +4
Diffusion language models (DLMs) have recently emerged as an alternative modeling paradigm to autoregressive LMs, offering advantages such as parallel generation and bidirectional…
Don't Solve, Just Compare: Tiny Advisors for Runtime Intervention in LLM Agents
Yanze Jiang, Mingxuan Li, Yuhao Wang +2
LLM agents are emerging as an important paradigm for real-world tasks that require reasoning, tool use, and sequential decision-making. As these agents operate over longer horizons…
Securing LLM Agents Need Intent-to-Execution Integrity
Wenjie Qu, Ming Xu, Peiran Wang +3
This position paper argues that securing LLM agents requires first defining an end-to-end correctness property that specifies when an agent's execution faithfully reflects the user…
BadDLM: Backdooring Diffusion Language Models with Diverse Targets
Shengfang Zhai, Xiaoyang Ji, Yuling Shi +6
Diffusion language models (DLMs) have recently emerged as an alternative modeling paradigm to autoregressive (AR) language models, enabling parallel generation and bidirectional co…
ClawMark: A Living-World Benchmark for Multi-Turn, Multi-Day, Multimodal Coworker Agents
Fanqing Meng, Lingxiao Du, Zijian Wu +46
Language-model agents are increasingly used as persistent coworkers that assist users across multiple working days. During such workflows, the surrounding environment may change in…
Purify Once, Edit Freely: Breaking Image Protections under Model Mismatch
Qichen Zhao, Shengfang Zhai, Xinjian Bai +4
Diffusion models enable high-fidelity image editing but can also be misused for unauthorized style imitation and harmful content generation. To mitigate these risks, proactive imag…
IMMACULATE: A Practical LLM Auditing Framework via Verifiable Computation
Yanpei Guo, Wenjie Qu, Linyu Wu +7
Commercial large language models are typically deployed as black-box API services, requiring users to trust providers to execute inference correctly and report token usage honestly…
MemPot: Defending Against Memory Extraction Attack with Optimized Honeypots
Yuhao Wang, Shengfang Zhai, Guanghao Jin +3
Large Language Model (LLM)-based agents employ external and internal memory systems to handle complex, goal-oriented tasks, yet this exposes them to severe extraction attacks, and…
DMark: Order-Agnostic Watermarking for Diffusion Large Language Models
Linyu Wu, Linhao Zhong, Wenjie Qu +5
Diffusion large language models (dLLMs) offer faster generation than autoregressive models while maintaining comparable quality, but existing watermarking methods fail on them due…
GuardReasoner-VL: Safeguarding VLMs via Reinforced Reasoning
Yue Liu, Shengfang Zhai, Mingzhe Du +9
To enhance the safety of VLMs, this paper introduces a novel reasoning-based VLM guard model dubbed GuardReasoner-VL. The core idea is to incentivize the guard model to deliberativ…
Silent Leaks: Implicit Knowledge Extraction Attack on RAG Systems through Benign Queries
Yuhao Wang, Wenjie Qu, Shengfang Zhai +5
Retrieval-Augmented Generation (RAG) systems enhance large language models (LLMs) by incorporating external knowledge bases, but this may expose them to extraction attacks, leading…
Life-Cycle Routing Vulnerabilities of LLM Router
Qiqi Lin, Xiaoyang Ji, Shengfang Zhai +4
Large language models (LLMs) have achieved remarkable success in natural language processing, yet their performance and computational costs vary significantly. LLM routers play a c…
Sparse Autoencoder as a Zero-Shot Classifier for Concept Erasing in Text-to-Image Diffusion Models
Zhihua Tian, Sirun Nan, Ming Xu +5
Text-to-image (T2I) diffusion models have achieved remarkable progress in generating high-quality images but also raise people's concerns about generating harmful or misleading con…
Efficient Input-level Backdoor Defense on Text-to-Image Synthesis via Neuron Activation Variation
Shengfang Zhai, Jiajun Li, Yue Liu +7
In recent years, text-to-image (T2I) diffusion models have gained significant attention for their ability to generate high quality images reflecting text prompts. However, their gr…
GuardReasoner: Towards Reasoning-based LLM Safeguards
Yue Liu, Hongcheng Gao, Shengfang Zhai +9
As LLMs increasingly impact safety-critical applications, ensuring their safety using guardrails remains a key challenge. This paper proposes GuardReasoner, a new safeguard for LLM…
Membership Inference on Text-to-Image Diffusion Models via Conditional Likelihood Discrepancy
Shengfang Zhai, Huanran Chen, Yinpeng Dong +5
Text-to-image diffusion models have achieved tremendous success in the field of controllable image generation, while also coming along with issues of privacy leakage and data copyr…
Discovering Universal Semantic Triggers for Text-to-Image Synthesis
Shengfang Zhai, Weilong Wang, Jiajun Li +3
Recently text-to-image models have gained widespread attention in the community due to their controllable and high-quality generation ability. However, the robustness of such model…
Text-to-Image Diffusion Models can be Easily Backdoored through Multimodal Data Poisoning
Shengfang Zhai, Yinpeng Dong, Qingni Shen +3
With the help of conditioning mechanisms, the state-of-the-art diffusion models have achieved tremendous success in guided image generation, particularly in text-to-image synthesis…
NCL: Textual Backdoor Defense Using Noise-augmented Contrastive Learning
Shengfang Zhai, Qingni Shen, Xiaoyi Chen +4
At present, backdoor attacks attract attention as they do great harm to deep learning models. The adversary poisons the training data making the model being injected with a backdoo…
Apple of Sodom: Hidden Backdoors in Superior Sentence Embeddings via Contrastive Learning
Xiaoyi Chen, Baisong Xin, Shengfang Zhai +3
This paper finds that contrastive learning can produce superior sentence embeddings for pre-trained models but is also vulnerable to backdoor attacks. We present the first backdoor…
Kallima: A Clean-label Framework for Textual Backdoor Attacks
Xiaoyi Chen, Yinpeng Dong, Zeyu Sun +3
Although Deep Neural Network (DNN) has led to unprecedented progress in various natural language processing (NLP) tasks, research shows that deep models are extremely vulnerable to…