I am a final year Ph.D. student in Computer Science at Purdue University. I am fortunate to work with Professor Xiangyu Zhang.

My research focuses on building secure and trustworthy foundations for modern generative and agentic AI systems (LLMs/VLMs/DMs/Agents) by mitigating adversarial threats such as backdoors, jailbreaks, and AIGC-based deception. I aim to endow these models with human-level risk awareness through a unified, scalable framework that integrates threat characterization, red-teaming, and awareness-oriented alignment training.

I serve as the team leader and core member of the Perspecta-PurdueUMass team, which competes in the TrojAI Program, an AI backdoor detection competition held by IARPA. Over the past four years, our team has achieved top-tier performance, securing leading positions in 14 out of 20 rounds. In the course of this competition, I have developed and refined a suite of scanning methodologies for detecting backdoors across a variety of machine learning models, including object detection systems, malware detectors, and large language models. I also participate in the Amazon Nova AI Challenge as the co-lead of the PurCL team, where we develope advanced red-teaming tools to help build trustworthy AI coding agents. Our team was awarded first place as the winning attacker team ($250,000) in the finals.

📢: I am always open to discussions and collaborations. If you are interested in exploring ideas related to AI safety and security, please feel free to contact me via email.

News

Dec. 2025: Two papers on VLM-based deepfake detection and backdoor mitigation got accepted to WACV 2026.

Nov. 2025: Our paper on AI-generated text origin detection got accepted to EMNLP 2025.

Nov. 2025: Three papers got accepted to ResponsibleFM Workshop@NeurIPS 2025.

Oct. 2025: Our paper on benchmarking diffusion models safety got accepted to ICCV 2025.

Jul. 2025: Our team PurCL won the first place in the Amazon Nova AI Challenge. Shout out to the team co-lead Xiangzhe and all teammates/advisors! BOILER UP!

Selected Publications [Full List] (* equal contribution)

• AuthGuard: Generalizable Deepfake Detection via Language Guidance
  Guangyu Shen, Zhihua Li, Xiang Xu, Tianchen Zhao, Zheng Zhang, Dongsheng An, Zhuowen Tu, Yifan Xing, Qin Zhang
  Winter Conference on Applications of Computer Vision 2026 (WACV 2026)
  

• From Poisoned to Aware: Fostering Backdoor Self-Awareness in LLMs
  Guangyu Shen, Siyuan Cheng, Xiangzhe Xu, Yuan Zhou, Hanxi Guo, Zhuo Zhang, Xiangyu Zhang
  NeurIPS 2025 Workshop on Socially Responsible and Trustworthy Foundation Models (ResponsibleFM)
  

• ASTRA: Autonomous Spatial-Temporal Red-teaming for AI Software Assistants
  Xiangzhe Xu^*, Guangyu Shen^*, Zian Su, Siyuan Cheng, Hanxi Guo, Lu Yan, Xuan Chen, Jiasheng Jiang, Xiaolong Jin, Chengpeng Wang, Zhuo Zhang, Xiangyu Zhang
  NeurIPS 2025 Workshop on Socially Responsible and Trustworthy Foundation Models (ResponsibleFM)
  🏆 Winning red-teaming solution in Amazon Nova AI Challenge
  

• Rapid Optimization for Jailbreaking LLMs via Subconscious Exploitation and Echopraxia
  Guangyu Shen^*, Siyuan Cheng^* Kaiyuan Zhang, Lu Yan, Shengwei An, Zhuo Zhang, Guanhong Tao, Shiqing Ma, Xiangyu Zhang
  NeurIPS 2025 Worws_self_awarenesskshop on Socially Responsible and Trustworthy Foundation Models (ResponsibleFM)
  

• BAIT: Large Language Model Backdoor Scanning by Inverting Attack Target
  Guangyu Shen^*, Siyuan Cheng^*, Zhuo Zhang, Guanhong Tao, Kaiyuan Zhang, Hanxi Guo, Lu Yan, Xiaolong Jin, Shengwei An, Shiqing Ma, Xiangyu Zhang
  Proceedings of the 46th IEEE Symposium on Security and Privacy (S&P 2025)
  

• ODSCAN: Backdoor Scanning for Object Detection Models
  Siyuan Cheng^*, Guangyu Shen^*, Guanhong Tao, Kaiyuan Zhang, Zhuo Zhang, Shengwei An, Xiangzhe Xu, Yingqi Liu, Shiqing Ma, Xiangyu Zhang
  Proceedings of the 45th IEEE Symposiums on Security and Privacy (S&P 2024)
  

• UNIT: Backdoor Mitigation via Automated Neural Distribution Tightening
  Siyuan Cheng^*, Guangyu Shen^*, Kaiyuan Zhang, Guanhong Tao, Shengwei An, Hanxi Guo, Shiqing Ma, Xiangyu Zhang
  The 18th European Conference on Computer Vision (ECCV 2024)
  

• Django: Detecting Trojans in Object Detection Models via Gaussian Focus Calibration
  Guangyu Shen^*, Siyuan Cheng^*, Guanhong Tao, Kaiyuan Zhang, Yingqi Liu, Shengwei An, Shiqing Ma, Xiangyu Zhang
  Proceedings of 37th Conference on Neural Information Processing Systems (NeurIPS 2023)
  

• PICCOLO: Exposing Complex Backdoors in NLP Transformer Models
  Yingqi Liu^*, Guangyu Shen^*, Guanhong Tao, Shengwei An, Shiqing Ma, Xiangyu Zhang Proceedings of the 43rd IEEE Symposiums on Security and Privacy (S&P 2022)
  

• Constrained Optimization with Dynamic Bound-scaling for Effective NLP Backdoor Defense
  Guangyu Shen^*, Yingqi Liu^*, Guanhong Tao, Qiuling Xu, Zhuo Zhang, Shengwei An, Shiqing Ma, Xiangyu Zhang
  Proceedings of the 39th International Conference on Machine Learning (ICML 2022)
  

• Complex Backdoor Detection by Symmetric Feature Differencing
  Yingqi Liu^*, Guangyu Shen^*, Guanhong Tao, Zhenting Wang, Shiqing Ma, Xiangyu Zhang
  IEEE/CVF Conference on Computer Vision and Pattern Recognition 2022 (CVPR 2022)
  

• Backdoor Scanning for Deep Neural Networks through K-Arm Optimization
  Guangyu Shen^*, Yingqi Liu^*, Guanhong Tao, Shengwei An, Qiuling Xu, Siyuan Cheng, Shiqing Ma, Xiangyu Zhang
  Proceedings of Thirty-eighth International Conference on Machine Learning (ICML 2021)
  

Awards & Honors

Fellowship

• Bilsland Dissertation Fellowship, Purdue, 2025

Competition Record

• 1st place for Amazon Nova AI Challenge ($250,000)
• 1st place for TrojAI 14 out of 20 rounds
• 2nd place for 2 tracks in Trojan Detection Competition (TDC2022)
  • Target Label Prediction
  • Trigger Synthesis
• 3nd place and most efficient method award for Track II: Backdoor Trigger Recovery for Models in The Competition for LLM and Agent Safety (CLAS2024)

Teaching

• Guest lecture at Virginia Tech, CS 6804: AI Security and Privacy, invited by Prof. Shengwei An, Nov. 2025
• Guest lecture at Rice University, COMP 620: Machine Learning System Seminar, invited by Prof. Yuke Wang, Nov. 2025
• Guest lecture at University of Uath, CS 6958: Machine Learning Security, invited by Prof. Guanhong Tao, Oct. 2025
• Guest lecture at University of Georgia, CSCI 8000: Red-Teaming for LLM Agents, invited by Prof. Zhen Xiang, Oct. 2025
• Guest lecture at University of Uath, CS 6958: Machine Learning Security, invited by Prof. Guanhong Tao, Oct. 2024
• Teaching Assistance at Purdue University, CS 59200 - AI and Security, Aug. 2024
• Guest lecture at University of Massachusetts, Amherst, COMPSCI 360: Introduction to Computer and Network Security, invited by Prof. Shiqing Ma, Feb. 2024

Services

Competition Co-chair

• IEEE Trojan Removal Competition, 2022

Program Committee

• ACM Conference on Computer and Communications Security (CCS): 2025, 2026
• 8th Deep Learning Security and Privacy Workshop(DLSP): 2025
• Workshop on Backdoors in Deep Learning: The Good, the Bad, and the Ugly(BUGS), NeurIPS 2023
• Workshop on Secure and Trustworthy Deep Learning Systems (SecTL), AsiaCCS 2023

Reviewer

• IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR): 2022,2023
• International Conference on Machine Learning (ICML): 2022,2023,2024,2025
• European Conference on Computer Vision (ECCV): 2022
• International Conference on Computer Vision (ICCV): 2023
• Conference on Neural Information Processing Systems (NeurIPS): 2022,2023
• International Conference on Learning Representations (ICLR): 2025

Experiences

• Applied Scientist Intern, Amazon AWS AI Lab, May.2024-Aug.2024, May.2023-Aug.2023
• Research Assistant, working with Prof.Baijian Yang, Purdue University, Aug.2019-Jan.2020
• Summer Research Intern, working with Prof.Junfeng Yang and Prof.Baishakhi Ray, Columbia University, May.2019-Aug.2019

Personal

I love movies and music. 🏂 is my new favorite sport. :p