| I am a PhD candidate in computer science at Purdue University, advised by Prof. Ninghui Li. I study security and privacy risks in machine learning and LLM agents, from both adversarial and defensive perspectives. My current research focuses on three directions: 🚨 AI Security & Safety Uncovering threats from LLM misuse 🔍 Machine Learning Privacy Assessing information leakage in ML models 🛡️ Data Privacy Identifying and protecting data privacy My research has been recognized and supported by the Ross Fellowship (2023–2027), Presidential Doctoral Excellence Award (2023–2027), and Herbold Scholarship (2023–2024). | Selected Publications - Automated Profile Inference with Language Model Agents
Yuntao Du, Zitao Li, Bolin Ding, Yaliang Li, Hanshen Xiao, Jingren Zhou, Ninghui Li
ACL 2026 (Findings) [pdf] [code]
AI Security & Safety First study showing that LLM agents enable automated doxing at web scale. - Beyond Data Privacy: New Privacy Risks for Large Language Models
Yuntao Du, Zitao Li, Ninghui Li, Bolin Ding
Data Eng. Bulletin 2025 [pdf]
AI Security & Safety Systematizes new privacy risks of LLM agents beyond training-data leakage. - Imitative Membership Inference Attack
Yuntao Du, Yuetian Chen, Hanshen Xiao, Bruno Ribeiro, Ninghui Li
USENIX Security 2026 [pdf] [code] [blog]
Machine Learning Privacy A new shadow training paradigm for MIAs with significantly reduced computation. - Cascading and Proxy Membership Inference Attacks
Yuntao Du, Jiacheng Li, Yuetian Chen, Kaiyuan Zhang, Zhizhen Yuan, Hanshen Xiao, Bruno Ribeiro, Ninghui Li
NDSS 2026 [pdf] [code] [blog]
Machine Learning Privacy Formulates and categorizes MIAs and first exploits membership dependencies. - Privacy Leakage from a Thousand Words: Sub-Pixel Location Recovery from Dot Maps
Yuntao Du*, Tanishq Praveen Pauskar*†, Hao Wang, Jing Su, Ninghui Li
Manuscript
Data Privacy First to identify re-identification risks of dot maps with 1-meter accuracy on national-scale maps. - Systematic Assessment of Tabular Data Synthesis
Yuntao Du, Ninghui Li
CCS 2025 [pdf] [code] [blog]
Data Privacy Proposes a unified evaluation framework for tabular data synthesis algorithms. See the full publication list → | Selected Awards & Honors - National Winner, Innovation Bowl (Purdue team lead), 2026 Coverage: [Purdue CS][Radiance]
- NDSS Fellowship, Internet Society (1 of 24 worldwide), 2026
- Ross Fellowship, Purdue University, 2023
- Herbold Scholarship, Purdue University (1 of 7), 2023
- Presidential Doctoral Excellence Award, Purdue University (1 of 150), 2023
- Excellent Masters Dissertation, China (1 of 43), 2023
- Provincial Outstanding Graduates, Zhejiang, China, 2023
- National Scholarship, China (0.1%), 2021-2022
| Service - Program Committee: VLDB (2027), AsiaCCS (2027), NeurIPS (2026), ICLR (2025-2026), AISTATS (2025-2026), WWW (2026), CODASPY (2026), WSDM (2026), CIKM (2024-2026), SIGIR (2023-2026), AAAI (2023-2027)
- Poster Program Committee: IEEE S&P (2026)
- Journal Reviewers: CSUR, TDSC, TOPS, VLDBJ, TKDE, TORS, TBD
| |
