Zichuan Li 李子川


zichuan7 AT illinois DOT edu
OR zichuan DOT li AT outlook DOT com

Welcome! I am a first/second/third/fourth-year Computer Science Ph.D. student at University of Illinois Urbana-Champaign advised by Prof. Luyi Xing. Previously, I received my B.Eng. and M.Eng. degree from Wuhan University.

My research focuses on the intersection of AI security and system security, e.g., identifying and mitigating security issues in emerging AI applications and systems, and leveraging AI techniques to help address long-standing challenges in traditional system and software security.
Github Twitter Mail Telegram Blog CV Calendar

News

2026-01-06 I received the Internet Society Fellowship to attend the NDSS Symposium!link
2025-09-12 I am happy to be recognized as a Noteworthy Reviewer in USENIX 2025!link
2025-08-02 We team Hoosiers won the πŸ… First place at the AgentX competiton hosted by Berkeley RDI.link
2025-02-10 We team Hoosiers (Jian and I) won πŸ₯ˆ Second Place in the Safety Track of the LLM Agent Hackathon hosted by Berkeley RDI.link
2024-08-23 Spent a wonderful summer with Binja folks! Check out our blog on Binja's UEFI plugin!link
2024-05-14 Phoenix acknowledged my report and published a vulnerability fixlink

Show more ↓


Industry Experience

2024.05 ~ 2024.08     Vector35 Inc.     Summer Intern
Built and imporved UEFI plugin for Binary Ninja.
Advanced UEFI Analysis with Binary Ninja, The Fallback Type Library
2022.06 ~ 2022.08     OPPO Telecom.     Summer Intern
Researched on Android clipboard privacy protection method.

Publications

2026

[5]What's in Your Agent's Context? Context Privilege Escalation Attacks against AI Agent Harness pdfdemobib
Zichuan Li, Jian Cui, Ashley Chen, Xiaojing Liao, Luyi Xing
arXiv:2609.01222
See our attack demos exploiting agent harness like Claude Code, Codex, Gemini and more!

[4]Characterizing the Ecosystem of Open-Source Uncensored Large Language Models: Supply Chain, Governance, and Risks pdfbib
Zilong Lin, Zichuan Li, Xiaojing Liao, XiaoFeng Wang
ACM Internet Measurement Conference (IMC), Karlsruhe, Germany, October 12-16, 2026

[3]Les Dissonances: Cross-Tool Harvesting and Polluting in Pool-of-Tools Empowered LLM Agents pdfcodebib
Zichuan Li*, Jian Cui*, Xiaojing Liao, Luyi Xing
Network and Distributed System Security Symposium (NDSS) 2026, San Diego, California, USA, February 24-27, 2026
Second place in LLM Agent Hackathon competition.

2025

[2]HarnessAgent: Scaling Automatic Fuzzing Harness Construction with Tool-Augmented LLM Pipelines pdfbib
Kang Yang, Yunhang Zhang, Zichuan Li, Guanhong Tao, Jun Xu, Xiaojing Liao
arXiv:2512.03420

[1]Safeguard-by-Development: A Privacy-Enhanced Development Paradigm for Multi-Agent Collaboration Systems pdfbib
Jian Cui, Zichuan Li, Luyi Xing, Xiaojing Liao
arXiv:2505.04799
First place in AgentX competition. Adopted by AG2.


Vulnerability Credits


Honors & Awards

01/2026 NDSS Fellowship
02/2025 2nd at the safety track of LLM Agent Hackathon
08/2025 1st at the safety track of AgentX
05/2023 Outstanding Graduate Student, Wuhan University.
12/2022 7th at Datacon Data Security Analytics Competition: IoT security Track. ($700)
09/2022 1st rank scholarship for graduate students at Wuhan University. ($600)
12/2021 1st at Datacon Data Security Analytics Competition: Email Security Track. ($7000)
09/2021 1st rank scholarship for graduate students at Wuhan University. ($600)
05/2021 1st rank scholarship for research excellent at Wuhan University
11/2020 3rd Prize at Coremail Email Security Competition. ($1500)
09/2019 3rd Prize at Chinese Information Hiding Competition.

Teaching & Services



Copyright Β© 2023-2026 Built by river-li | Last updated on 2026-09-03.

×
Bibliography Entry