Publications

Preprint
AgentLens: Interpretable Safety Steering via Mechanistic Subspaces for Multi-Turn Coding Agent

AgentLens: Interpretable Safety Steering via Mechanistic Subspaces for Multi-Turn Coding Agent

Weidi Luo, Qiming Zhang, Yihao Quan, Mingyu Jin, Jie Cai, Chaowei Xiao, Jingcheng Niu, Zhen Xiang

arXivGitHub
ICLR 2026
Doxing via the Lens: Revealing Location-related Privacy Leakage on Multi-modal Large Reasoning Models

Doxing via the Lens: Revealing Location-related Privacy Leakage on Multi-modal Large Reasoning Models

Weidi Luo*, Tianyu Lu*, Qiming Zhang*, Xiaogeng Liu, Bin Hu, Yue Zhao, Jieyu Zhao, Song Gao, Patrick McDaniel, Zhen Xiang, Chaowei Xiao

* denotes equal contribution

International Conference on Learning Representations (ICLR), 2026

🌐Project PagePDFarXivGitHub🤗Dataset
Preprint
Code Agent can be an End-to-end System Hacker: Benchmarking Real-world Threats of Computer-use Agent

Code Agent can be an End-to-end System Hacker: Benchmarking Real-world Threats of Computer-use Agent

Weidi Luo*, Qiming Zhang*, Tianyu Lu, Xiaogeng Liu, Bin Hu, Hung-Chun Chiu, Siyuan Ma, Yizhe Zhang, Xusheng Xiao, Yinzhi Cao, Zhen Xiang, Chaowei Xiao

* denotes equal contribution

Oral - ICML 2026 Workshop on Agents in the Wild

🌐Project PagePDFarXivGitHub🤗Dataset