About Me

I am a Security Researcher at Ant Group. Before joining Ant Group, I was a Researcher at Huawei 2012 Labs.

My research focuses on AI agent security, particularly the discovery and analysis of vulnerabilities in autonomous agents. I am especially interested in the following topics:

• Adaptive attacks against autonomous agents
• System-level vulnerability discovery

I investigate how attacks unfold during task execution and interactions with tools and the environment, with a focus on vulnerabilities that emerge at runtime and across system components. My earlier work explored adversarial and jailbreaking attacks against large language models, as well as LLM-based code vulnerability detection.

I received my Ph.D. from Huazhong University of Science and Technology (HUST) in June 2025, supervised by Prof. Ming Wen. Previously, I received my master’s degree in Software Engineering from National University of Defense Technology in 2021, supervised by Prof. Jie Yu. I have published 10+ papers in leading conferences and journals, including ACL, EMNLP, AAAI, and FSE. In addition, I serve as a program committee member or reviewer for conferences including ICLR, AAAI, KDD, ACM MM, and EMNLP, and as a reviewer for journals including IEEE TDSC and ACM TALLIP.

I welcome discussions and collaborations on agent security, adaptive attacks, and vulnerability discovery. Please feel free to contact me by email: xiaohudu at outlook.com or duxiaohu.dxh at antgroup.com