7 papers
Staying on the Attack Path: Structured State for Long-Horizon Automated Penetration Testing
Weizhe Wang, Yitong Zhang, Yao Zhang +4
Large language model (LLM) based agents are increasingly applied to cybersecurity tasks such as vulnerability discovery and automated penetration testing. On long-horizon security…
Domain Decoupling Attack: Exploiting the Validation Gap Between Protective DNS and Shared Edge Routing
Weizhe Wang, Minhong Dong, Jinhao Li +8
Command-and-control, data transfer, and censorship circumvention can hide in traffic to widely used services. DNS-bound access control grants permits based on resolved IP addresses…
SFCoT: Safer Chain-of-Thought via Active Safety Evaluation and Calibration
Yu Pan, Wenlong Yu, Tiejun Wu +4
Large language models (LLMs) have demonstrated remarkable capabilities in complex reasoning tasks. However, they remain highly susceptible to jailbreak attacks that undermine their…
Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale
Yi Liu, Weizhe Wang, Ruitao Feng +5
The rise of AI agent frameworks has introduced agent skills, modular packages containing instructions and executable code that dynamically extend agent capabilities. While this arc…
PolyJailbreak: Cross-Modal Jailbreaking Attacks on Black-Box Multimodal LLMs
Xinkai Wang, Beibei Li, Zerui Shao +3
Multimodal large language models (MLLMs) have become integral to a wide range of real-world applications by jointly reasoning over text and visual inputs. However, despite recent a…
AndroTruth: A Reliable Benchmark Android Malware Dataset Derived from Technical Expert Reports
Hongpeng Bai, Yao Zhang, Minhong Dong +6
Reliable family labels are essential for Android malware analysis, yet most widely used benchmarks derive such labels from aggregated VirusTotal engine outputs. Because antivirus v…
VulnRepairEval: An Exploit-Based Evaluation Framework for Assessing Large Language Model Vulnerability Repair Capabilities
Weizhe Wang, Wei Ma, Qiang Hu +6
The adoption of Large Language Models (LLMs) for automated software vulnerability patching has shown promising outcomes on carefully curated evaluation sets. Nevertheless, existing…