Xiang Zheng

Research Assistant Professor
Hong Kong Institute of AI for Science (HKAI-Sci)
City University of Hong Kong
Chief Scientist, SciencePal 2.0
Google Scholar | GitHub | CityUHK Scholar

I am a Research Assistant Professor at the Hong Kong Institute of AI for Science (HKAI-Sci), City University of Hong Kong. I also work closely with Prof. Xingjun Ma at the Institute of Trustworthy Embodied AI, Fudan University. My research develops robust and efficient RL algorithms for trustworthy science, computer-use, and embodied agents. I also collaborate closely with industry: my research has contributed to ByteDance’s AgentArmor agent-safety framework, Foxconn Research Institute’s FoxBrain foundation model (in collaboration with NVIDIA), and HKAI-Sci’s SciencePal 2.0 self-evolving science agent. I have been honored to receive the prestigious CityUHK Presidential PhD Scholarship.

I received my Ph.D. from the Department of Computer Science at City University of Hong Kong in 2024, under the guidance of Prof. Cong Wang. Prior to that, I earned my Master’s degree in Control Science and Engineering from the Department of Automation at Tsinghua University in 2019, where I was advised by Prof. Tao Zhang. I hold dual Bachelor’s degrees in Automation and Mathematics from the Shen Yuan Honors College (formerly the School of Advanced Engineering) at Beihang University, where I graduated in 2016.

I currently have 1–2 openings for Research Assistants (RAs) and Ph.D. students, co-supervised with Prof. Wei-Ying Ma, on AI for Science and Recursive Self-Improving Agents. Interested candidates are welcome to email me at xiang.zheng@cityu.edu.hk with a CV and a brief research statement.

Latest News

Jun 16, 2026 Giving an invited talk at the 2026 CityUDG: “Hands-on with Claude Code: Frontier Code Agents for Research and Browser/Computer Use”.
Jun 05, 2026 Invited to speak at the 2026 Tencent Cloud AI Industry Applications Summit (China National Convention Center, Beijing) on “SciencePal 2.0: Your Self-Evolving Science Agent Team”.
Jun 03, 2026 Invited as a reviewer for ACM Computing Surveys (CSUR).
May 23, 2026 Invited as a reviewer for IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI).
May 14, 2026 Received the ICML 2026 Gold Reviewer Award (top 25% of reviewers).
May 05, 2026 Our Defense-to-Attack jailbreak study on VLMs is accepted by Pattern Recognition.
May 01, 2026 Two papers accepted to ICML’26: Just Ask (curious code agents revealing system prompts in frontier LLMs) and STARE (step-wise temporal red-teaming of multi-modal toxicity).
Mar 31, 2026 We release System-Prompt-Open, an open database of system prompts extracted from frontier LLMs. Check out the project website and GitHub repo.
Mar 28, 2026 Our survey on embodied AI safety with 400+ papers is now available on arXiv, covering risks, attacks, and defenses across perception, cognition, planning, interaction, and agentic systems.
Mar 09, 2026 Joined HKAI-Sci as Research Assistant Professor.
Mar 04, 2026 Our OpenRedRL benchmark for RL-based red teaming is published in Frontiers of Computer Science.
Feb 20, 2026 Our work on red teaming text-to-image generators is accepted by CVPR’26.
Jan 29, 2026 Released JustAsk, a framework where curious code agents reveal system prompts in frontier LLMs.
Jan 15, 2026 Our survey on large model and agent safety is published in Foundations and Trends® in Privacy and Security.
Jan 23, 2025 Our work on reinforced defense for VLMs is accepted by ICLR’25.
Dec 14, 2024 Our work on RL-based auditing for LLMs is accepted by AAAI’25.
Apr 17, 2024 Our work on intrinsic motivation for RL is accepted by IJCAI’24.
Mar 22, 2024 Our work on adversarial policy learning in RL is accepted by DSN’24.

Selected Publications

  1. ICML’26
    STARE: Step-wise Temporal Alignment and Red-teaming Engine for Multi-modal Toxicity Attack
    Xutao Mao, Liangjie Zhao, Tao Liu, Xiang Zheng†, Hongying Zan, and Cong Wang†
    In International Conference on Machine Learning (ICML), 2026
    arXiv
  2. ICML’26
    Just Ask: Curious Code Agents Reveal System Prompts in Frontier LLMs
    Xiang Zheng, Yutao Wu, Hanxun Huang, Yige Li, Xingjun Ma†, Bo Li, Yu-Gang Jiang, and Cong Wang†
    In International Conference on Machine Learning (ICML), 2026
    arXiv Demo GitHub Website
  3. PR
    Defense-to-attack: Bypassing weak defenses enables stronger jailbreaks in Vision-Language Models
    Yunhan Zhao, Xiang Zheng, Yige Li, and Xingjun Ma†
    Pattern Recognition, 2026
    DOI arXiv
  4. Preprint
    Safety in Embodied AI: A Survey of Risks, Attacks, and Defenses
    Xiao Li*, Xiang Zheng*, Yifeng Gao, Xinyu Xia, Yixu Wang, Xin Wang, Ye Sun, Yunhan Zhao, Ming Wen, Jiayu Li, Zixing Chen, Xun Gong, Yi Liu, Yige Li, Yutao Wu, Cong Wang, Jun Sun, Yixin Cao, Zhineng Chen, Jingjing Chen, Tao Gui, Qi Zhang, Zuxuan Wu, Xipeng Qiu, Xuanjing Huang, Tiehua Zhang, Zhipeng Wei, Kun Wang, Xinfeng Li, Hanxun Huang, Sarah Erfani, James Bailey, Jianping Wang, Chaowei Xiao, Ran He, Bo Li, Xingjun Ma†, and Yu-Gang Jiang†
    arXiv preprint, 2026
    arXiv GitHub Website
  5. TSC
    Adapting Large Language Models for Encrypted Traffic Analysis Services: An Efficient Realization with Mixture of LoRA Experts
    Yi Liu, Xiang Zheng, Chengjun Cai, Xingliang Yuan, and Cong Wang†
    IEEE Transactions on Services Computing, 2026
    DOI
  6. FCS
    OpenRedRL: A Light-Weight Benchmark for Reinforcement Learning-Based Red Teaming
    Xiang Zheng, Xingjun Ma, Wei-Bin Lee, and Cong Wang†
    Frontiers of Computer Science, 2026
    DOI arXiv
  7. CVPR’26
    GenBreak: Red Teaming Text-to-Image Generators Using Large Language Models
    Zilong Wang, Xiang Zheng†, Xiaosen Wang, Bo Wang, Xingjun Ma†, and Yu-Gang Jiang
    In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026
    arXiv
  8. FnT P&S
    Safety at Scale: A Comprehensive Survey of Large Model and Agent Safety
    Xingjun Ma, Yifeng Gao, Yixu Wang, Ruofan Wang, Xin Wang, Ye Sun, Yifan Ding, Hengyuan Xu, Yunhao Chen, Yunhan Zhao, Hanxun Huang, Yige Li, Yutao Wu, Jiaming Zhang, Xiang Zheng, Yang Bai, Zuxuan Wu, Xipeng Qiu, Jingfeng Zhang, Yiming Li, Xudong Han, Haonan Li, Jun Sun, Cong Wang, Jindong Gu, Baoyuan Wu, Siheng Chen, Tianwei Zhang, Yang Liu, Mingming Gong, Tongliang Liu, Shirui Pan, Cihang Xie, Tianyu Pang, Yinpeng Dong, Ruoxi Jia, Yang Zhang, Shiqing Ma, Xiangyu Zhang, Neil Gong, Chaowei Xiao, Sarah Erfani, Tim Baldwin, Bo Li, Masashi Sugiyama, Dacheng Tao, James Bailey, and Yu-Gang Jiang
    Foundations and Trends in Privacy and Security, 2026
    arXiv
  9. ICLR’25
    BlueSuffix: Reinforced Blue Teaming for Vision-Language Models Against Jailbreak Attacks
    Yunhan Zhao, Xiang Zheng†, Lin Luo, Yige Li, Xingjun Ma†, and Yu-Gang Jiang
    In International Conference on Learning Representations (ICLR), 2025
    arXiv GitHub
  10. AAAI’25
    CALM: Curiosity-Driven Auditing for Large Language Models
    Xiang Zheng, Longxiang Wang, Yi Liu, Xingjun Ma, Chao Shen, and Cong Wang†
    In AAAI Conference on Artificial Intelligence (AAAI), 2025
    arXiv GitHub
  11. IJCAI’24
    Constrained intrinsic motivation for reinforcement learning
    Xiang Zheng, Xingjun Ma, Chao Shen, and Cong Wang†
    In International Joint Conference on Artificial Intelligence (IJCAI), 2024
    arXiv GitHub
  12. DSN’24
    Toward evaluating robustness of reinforcement learning with adversarial policy
    Xiang Zheng, Xingjun Ma, Shengjie Wang, Xinyu Wang, Chao Shen, and Cong Wang†
    In Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN), 2024
    arXiv GitHub
  13. AST
    Reinforcement learning with prior policy guidance for motion planning of dual-arm free-floating space robot
    Yuxue Cao, Shengjie Wang, Xiang Zheng, Wenke Ma, Xinru Xie, and Lei Liu
    Aerospace Science and Technology (AST), 2023
  14. RA-L
    Collision-free trajectory planning for a 6-DoF free-floating space robot via hierarchical decoupling optimization
    Shengjie Wang, Yuxue Cao, Xiang Zheng, and Tao Zhang
    IEEE Robotics and Automation Letters (RA-L), 2022
  15. IROS’21
    A multi-target trajectory planning of a 6-DoF free-floating space robot via reinforcement learning
    Shengjie Wang, Xiang Zheng, Yuxue Cao, and Tao Zhang
    In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2021
  16. CVPR’20
    Clean-label backdoor attacks on video recognition models
    Shihao Zhao, Xingjun Ma, Xiang Zheng, James Bailey, Jingjing Chen, and Yu-Gang Jiang
    In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020

Selected Awards and Honors

  • IJCAI Travel Grant, IJCAI Organization, 2024
  • Conference Grant, City University of Hong Kong, 2024
  • DSN Student Travel Grant, DSN Student Travel Awards Committee, 2024
  • Research Activities Fund, City University of Hong Kong, 2022
  • Institutional Research Tuition Grant, City University of Hong Kong, 2020
  • CityUHK Presidential PhD Scholarship (HK$1.56m for world-class PhD candidates), City University of Hong Kong, 2020
  • NII MOU Research Activities Grant, National Institute of Informatics, Japan, 2018
  • CSC Scholarship, China Scholarship Council, 2016
  • Stars of Advanced Engineering, the highest honor at the Shen Yuan Honors College, Beihang University, 2015

Professional Activities

Program Committee Member
  • NeurIPS 2026
  • ICML 2026
  • ICLR 2025, 2026
  • AAAI 2025, 2026
  • MM 2025, 2026
  • CVPR 2026
  • ECCV 2026
  • ICRA 2026
Journal Reviewer
  • ACM Computing Surveys (CSUR)
  • IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
  • IEEE Transactions on Dependable and Secure Computing (TDSC)
  • IEEE Transactions on Services Computing (TSC)
  • IEEE Transactions on Computers (TC)
External Reviewer
  • ACL 2026
  • NeurIPS 2025
  • ICNP 2025
  • ESORICS 2022
  • AsiaCCS 2022
  • RAID 2021
  • IEEE Internet of Things Journal (IoT-J)

Research & Work Experience

Visiting Researcher @ Xi'an Jiaotong University Senior Researcher @ Huawei
  • May 2020 - Aug 2020, Xi'an, China
Visiting Researcher @ Xi'an Jiaotong University Research Intern @ National Institute of Informatics Research Intern @ The University of New South Wales

Teaching Assistant

  • CS5293, Topics on Information Security, Semester B 2023/24, Code for Tutorial
  • CS4394, Information Security and Management, Semester A 2023/24
  • CS4293 / CS5293, Topics in Cybersecurity / Topics on Information Security, Semester B 2021/22
  • CS4394 / CS5294, Information Security and Management, Semester A 2021/22
  • CS4293 / CS6290, Topics in Cybersecurity / Privacy-enhancing Technologies, Semester B, 2020/21
  • CS2310 / CS2311, Computer Programming, Semester A, 2020/21