I am a year-1 PhD student in Computer Science Engineering at Hong Kong University of Science and Technology (HKUST), advised by Prof. Yangqiu Song. Currently, I am working with Dr. Haoran Li on LLM safety and privacy. During my undergraduate studies, I had the fortune of working with Prof. Minhao Cheng on backdoor attack on object detection models.

If you are interested in my works or would like to collaborate with me, please feel free to contact me via email (whuak at connect dot ust dot hk).

This webpage is last updated on 2025/09/29.

Research Interests

My research interest lies on the intersection of LLM reasoning and safety/privacy. Currently, I am working on:

• LLM Safety through the Lens of Compliance: Compliant Reasoner.
• LLM Contextual Integrity Reasoning: Context Reasoner and PrivaCI-Bench.
• Trustworthy Agent Protocol with Top-down Overseeing: MCIP.

Education

• Ph.D. in Computer Science at Hong Kong University of Science and Technology, 2025.02 - Now
• B.Eng. in Computer Science at Hong Kong University of Science and Technology, 2020.09 - 2024.06
• Exchange Semester in Technical University of Munich, 2023.04 - 2023.08

Working Experience

• Research Assistant at Hong Kong University of Science and Technology, 2024.07 - 2025.01

Publication

• Context Reasoner: Incentivizing Reasoning Capability for Contextualized Privacy and Safety Compliance via Reinforcement Learning. 2025.
  Wenbin Hu, Haoran Li, Huihao Jing, Qi Hu, Ziqian Zeng, Sirui Han, Heli Xu, Tianshu Chu, Peizhao Hu, Yangqiu Song.
  EMNLP 2025 (Main). [paper]

• MCIP: Protecting MCP Safety via Model Contextual Integrity Protocol. 2025.
  Huihao Jing, Haoran Li, Wenbin Hu, Qi Hu, Heli Xu, Tianshu Chu, Peizhao Hu, Yangqiu Song.
  EMNLP 2025 (Main). [paper]

• PrivaCI-Bench: Evaluating Privacy with Contextual Integrity and Legal Compliance. 2025.
  Haoran Li*, Wenbin Hu*, Huihao Jing*, Yulin Chen, Qi Hu, Sirui Han, Tianshu Chu, Peizhao Hu, Yangqiu Song.
  ACL 2025 (Main). [paper]

• Mitigating the Alignment Tax of RLHF. 2024.
  Yong Lin*, Hangyu Lin*, Wei Xiong*, Shizhe Diao*, Jianmeng Liu, Jipeng Zhang, Rui Pan, Haoxiang Wang, Wenbin Hu, Hanning Zhang, Hanze Dong, Renjie Pi, Han Zhao, Nan Jiang, Heng Ji, Yuan Yao, Tong Zhang.
  EMNLP 2024 (Main). [paper]

Preprints

• Safety Compliance: Rethinking LLM Safety Reasoning through the Lens of Compliance. 2025.09. Wenbin Hu, Huihao Jing, Haochen Shi, Haoran Li, Yangqiu Song. [paper]

• Node Level Graph Autoencoder: Unified Pretraining for Textual Graph Learning. 2024.08
  Wenbin Hu*, Huihao Jing*, Qi Hu*, Haoran Li, Yangqiu Song. [paper]

• Attacking by Aligning: Clean-Label Backdoor Attacks on Object Detection. 2023.
  Yize Cheng*, Wenbin Hu*, Minhao Cheng. [paper]

(* represents equal contribution.)

Awards and Honors

• Full Scholarship for Ph.D. Study
• Dean’s List (Term GPA > 3.7, top 10 percentile)

Main Courses

• AI & ML Courses: Machine Learning, Deep Learning, Computer Vision and Graphics, Natural Language Processing, Reinforcement Learning, Probabilistic Graph Model.
• CS Foundation Courses: C++, OOP, Computer Architecture, Operation System, Algorithm, Software Engineering.
• Data Mining: Knowledge Discovery and Data Mining, Machine Learning with Structured Data, Web Search Engine.
• MATH Courses: Multivariable Calculus, Linear Algebra, Abstract Algebra, Probability Theory, Convex Optimization, Mathematic Analysis, Ordinary Differential Equation.

Misc.

I have a keen interest in implementing popular methods from scratch. I have developed RLHF (PPO), GPT-2 (with causal transformer layers), Mixture-of-Expert, and etc. [code]