Wenxuan Bao
Staff Research Scientist
Research areas: Trustworthy AI, Privacy-Preserving Machine Learning, Federated Learning, Differential Privacy, LLM Security
Biography
Wenxuan Bao joined Visa Research as a Staff Research Scientist in February 2026, where he is a member of the GATE (GenAI Trust & Efficiency) team. He received his Ph.D. in Computer Science from the University of Florida in December 2025, advised by Dr. Vincent Bindschaedler, and earned his M.S. in Electrical and Computer Engineering from the same university in 2021. Prior to joining Visa full time, he was a research intern at Visa Research in summer 2024 and at NEC Laboratories America in summer 2022.
Wenxuan's research centers on privacy-preserving machine learning and the security, privacy, and reliability of AI agents. His work on differential privacy for deep learning includes data augmentation techniques as well as studies of how model architecture and feature selection affect differentially private learning and of the reliability and generalizability of differentially private machine learning. He has also developed plausible-deniability methods for deep learning and storage systems, and worked on provably secure covert messaging with diffusion models and inference attacks against speaker anonymization. His research has appeared at top-tier venues including NeurIPS, ACSAC, and SaTML. At Visa Research's GATE team, he develops algorithms to improve the safety, privacy, and reliability of AI agent systems. He serves as a reviewer for NeurIPS, ICLR, ICML, and AISTATS, and was named a NeurIPS Top Reviewer in 2025.
Publications
- Wenxuan Bao, Shan Jin, Hadi Abdullah, Anderson C. A. Nascimento, Vincent Bindschaedler, and Yiwei Cai. “Deep Learning with Plausible Deniability.” In the Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS 2025), 2025.
- Weidong Zhu, Wenxuan Bao, Vincent Bindschaedler, Sara Rampazzi, and Kevin R. B. Butler. “Enabling Plausible Deniability in Flash-based Storage through Data Permutation.” In the 41st Annual Computer Security Applications Conference (ACSAC 2025), 2025.
- Luke A. Bauer, Wenxuan Bao, and Vincent Bindschaedler. “Provably Secure Covert Messaging Using Image-based Diffusion Processes.” In the 3rd IEEE Conference on Secure and Trustworthy Machine Learning (SaTML 2025), 2025.
- Luke A. Bauer, Wenxuan Bao, Malvika Ranjitsinh Jadhav, and Vincent Bindschaedler. “Inference Attacks for X-Vector Speaker Anonymization.” In 2025 IEEE Security and Privacy Workshops (SPW 2025), 2025.
- Malvika Ranjitsinh Jadhav, Wenxuan Bao, and Vincent Bindschaedler. “Uncovering the Deceptive Tactics of Stalkerware: A Large Scale Measurement Study.” In the 21st ACM ASIA Conference on Computer and Communications Security (ACM ASIACCS 2026), 2026.
- Wenxuan Bao and Vincent Bindschaedler. “R+R: Towards Reliable and Generalizable Differentially Private Machine Learning.” In the 40th Annual Computer Security Applications Conference (ACSAC 2024), 2024.
- Wenxuan Bao, Francesco Pittaluga, Vijay Kumar B G, and Vincent Bindschaedler. “DP-Mix: Mixup-based Data Augmentation for Differentially Private Learning.” In the Thirty-Seventh Annual Conference on Neural Information Processing Systems (NeurIPS 2023), 2023.
- Wenxuan Bao, Luke A. Bauer, and Vincent Bindschaedler. “On the Importance of Architecture and Feature Selection in Differentially Private Machine Learning.” arXiv preprint arXiv:2205.06720, 2022.