I am a third-year PhD student at The Ohio State University (OSU), advised by Prof. Huan Sun and working closely with Prof. Yu Su. Previously, I obtained my Master's and Bachelor's degrees from University of Science and Technology of China (USTC), advised by Prof. Enhong Chen.
I am broadly interested in NLP and related fields. My current research focuses on language agents, with the goal of building trustworthy AI agents that can reliably assist humans by automating complex or tedious tasks. Recently, my work has explored agent safety and security (SkillHarm, MisAct, AutoElicit), model-based planning for web agents (WebDreamer) and evaluation of agents on long-horizon web search tasks (Mind2Web 2).
Feel free to reach out if you are interested in my research or potential collaborations.
News
-
⚔️SkillHarm has been accepted to EMNLP 2026 Main!
-
Excited to release ⚔️SkillHarm, our latest work on lifecycle-aware skill-based attacks via automated construction!
-
Excited to start my summer internship at Apple in Seattle 🏔️, working with Eldon Schoop on personalized GUI agents!
-
Our work on Misaligned Action Detection and Unintended Behavior Elicitation in CUAs are accepted to ICML 2026!
-
Excited to share our latest work on agent safety about Misaligned Actions and Unintended Behaviors in computer-use agents!
Selected Publications
All publications →Full list on my Google Scholar. (* indicates equal contribution.)