Portrait of Xiang Zheng

Xiang Zheng郑翔

Research Assistant Professor
Hong Kong Institute of AI for Science
City University of Hong Kong
Chief Scientist of SciencePal 2.0

About

I am a Research Assistant Professor at the Hong Kong Institute of AI for Science (HKAI-Sci), City University of Hong Kong, and work closely with Prof. Xingjun Ma at the Institute of Trustworthy Embodied AI, Fudan University. My research develops robust and efficient reinforcement-learning 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 (with NVIDIA), and HKAI-Sci’s SciencePal 2.0 self-evolving science agent. I was honored to receive the CityUHK Presidential PhD Scholarship.

I received my Ph.D. in Computer Science from City University of Hong Kong in 2024, advised by Prof. Cong Wang; my M.S. in Control Science and Engineering from Tsinghua University in 2019, advised by Prof. Tao Zhang; and dual B.S. degrees in Automation and Mathematics from the Shen Yuan Honors College, Beihang University, in 2016.

Research

AI for Science

Recursive Self-Improving Agent Team for Scientific Discovery

Trustworthy AI

Reinforcement Learning for Agent Safety

Embodied AI

Robust Robot Learning and VLA Safety

Recursive Self-Improvement

Safety of Self-Improving Agents

Openings

I currently have 1–2 openings for self-motivated Research Assistants 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.

News

2026.10
Invited to give talks, “SciencePal 2.0 Self-Evolving Research Agent”, at the Auto-Research Forum of CNCC 2026 in Chengdu and the Medical Large Model Agent Forum of CHIP 2026 in Dalian.
2026.09
Our recent work has been accepted to NeurIPS’26 (Internal Safety Collapse and VEX-Bench) and Findings of EMNLP’26 (HarmProfile). Congratulations to all collaborators!
2026.09
Invited to give a talk, “Trustworthy Recursive Self-Improvement”, at the Southern University of Science and Technology.
2026.08
Invited to give a talk, “Towards Trustworthy Agents: From Red Teaming to Trustworthy Recursive Self-Improving”, at the Victoria Harbour Academic Salon of Huawei Hong Kong Research Institute.
2026.06
DropVLA, an action-level backdoor attack on vision-language-action models, is accepted to IROS’26. Congratulations to Zonghuan and all collaborators!
2026.06
Invited to give a talk, “Hands-on with Claude Code: Frontier Code Agents for Research and Browser/Computer Use”, at the 2026 CityUHK Joint Summer School on AI for Science.
2026.06
Invited to give a talk, “SciencePal 2.0: Your Self-Evolving Science Agent Team”, at the 2026 Tencent Cloud AI Industry Applications Summit in Beijing.
2026.06
Invited to review for ACM Computing Surveys (CSUR) and IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI).
All news →

Publications

* equal contribution  † corresponding author

VEX-Bench: Benchmarking Verification Complexity of LLM-Generated Misinformation
Hanxun Huang, Yutao Wu, Qizhou Wang, Silvia Montaña-Niño, Yige Li, Xiang Zheng, Elif Buse Doyuran, Phoebe Matich,
NeurIPS 2026OpenReview
Internal Safety Collapse in Frontier Large Language Models
Yutao Wu, Xiao Liu, Hanxun Huang, Yige Li, Xiang Zheng, Yifeng Gao, Cong Wang, Bo Li, Xingjun Ma, Yu-Gang Jiang
NeurIPS 2026arXivcode
HarmProfile: Characterizing Harmful Distributions in Frontier LLMs
Zhouyuan Ma, Yutao Wu, Hanxun Huang, Xiang Zheng, Xiao Liu, Yixin Cao, Zuxuan Wu, Xingjun Ma, Yu-Gang Jiang
Findings of EMNLP 2026arXiv
Practice Makes Unsafe: Skill Misevolution in Self-Improving LLM Agents
Xutao Mao, Liangjie Zhao, Xiang Zheng†, Cong Wang†
arXiv:2608.12851arXivcode
Agent Hacks Agent: Autoresearch for Production-Agent Red-Teaming
Xutao Mao, Xiang Zheng†, Cong Wang†
arXiv:2607.11698arXiv
STARE: Step-wise Temporal Alignment and Red-teaming Engine for Multi-modal Toxicity Attack
Xutao Mao, Liangjie Zhao, Tao Liu, Xiang Zheng†, Hongying Zan, Cong Wang†
ICML 2026arXiv
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, Cong Wang†
ICML 2026arXivcodeproject
Defense-to-attack: Bypassing weak defenses enables stronger jailbreaks in Vision-Language Models
Yunhan Zhao, Xiang Zheng, Yige Li, Xingjun Ma†
Pattern Recognition (PR) 2026arXiv
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,
arXiv:2605.02900arXivcodeproject
OpenRedRL: A Light-Weight Benchmark for Reinforcement Learning-Based Red Teaming
Xiang Zheng, Xingjun Ma, Wei-Bin Lee, Cong Wang†
Frontiers of Computer Science (FCS) 2026arXiv
All publications →

Talks

2026.10
SciencePal 2.0 Self-Evolving Research Agent
Medical Large Model Agent Forum at CHIP 2026Dalian
2026.10
SciencePal 2.0 Self-Evolving Research Agent
Auto-Research Forum at CNCC 2026Chengdu
2026.09
Trustworthy Recursive Self-Improvement
Southern University of Science and TechnologyOnline
2026.08
Towards Trustworthy Agents: From Red Teaming to Trustworthy Recursive Self-Improving
Victoria Harbour Academic Salon at Huawei Hong Kong Research InstituteHong Kong
2026.06
Hands-on with Claude Code: Frontier Code Agents for Research and Browser/Computer Use
CityUHK Joint Summer School on AI for ScienceHong Kong
All talks →

Experience

2026–now
Research Assistant Professor at City University of Hong Kong (HKAI-Sci)
2024–2026
Postdoctoral Fellow at City University of Hong Kong (with Cong Wang)
2020–2024
Graduate Research Assistant at City University of Hong Kong (with Cong Wang)
2022–2023
Visiting Researcher at Xi'an Jiaotong University (with Chao Shen)
2019–2020
Visiting Researcher at Xi'an Jiaotong University (with Chao Shen and Xingjun Ma)
2018
Research Intern at National Institute of Informatics (with Tetsunari Inamura)
2016–2019
Graduate Research Assistant at Tsinghua University (with Tao Zhang)
2016
Research Intern at The University of New South Wales (with Elias Aboutanios)

Honors

2026
ICML Gold Reviewer Awardtop 25%
2024
IJCAI Travel GrantIJCAI Organization and AIJ Division
International Conference GrantCity University of Hong Kong
DSN Student Travel GrantDSN Student Travel Awards Committee
2022–2023
Research Activities FundCity University of Hong Kong
2020–2024
Institutional Research Tuition GrantCity University of Hong Kong
CityUHK Presidential PhD Scholarship2/73
2018
NII MOU Research Activities GrantNational Institute of Informatics Japan
2016
CSC ScholarshipChina Scholarship Council
2015
Stars of Advanced EngineeringShen Yuan Honors College at Beihang Universityhighest honor at the Honors College

Service

Organizer

Program Committee

  • NeurIPS 2026
  • ICML 2026
  • ICLR 2025–2027
  • CVPR 2026
  • ECCV 2026
  • EMNLP 2026
  • AAAI 2025–2027
  • ACM MM 2025–2026
  • 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)
  • Transactions on Machine Learning Research (TMLR)
  • IEEE Internet of Things Journal (IoT-J)

External Reviewer

  • ACL 2026
  • NeurIPS 2025
  • ICNP 2025
  • ESORICS 2022
  • AsiaCCS 2022
  • RAID 2021