Dung (Judy) Nguyen

Contact dung.t.nguyen (at) vanderbilt.edu.

prof_pic.jpg

I’m Judy | CS Ph.D. student @ Vanderbilt, advised by Prof. Taylor T. Johnson & Prof. Kevin Leach | SoICT alum | VinUni research alum | Vietnamese blood in me. I was lucky to spend the past two summers at the Center for AI Safety and Microsoft Research.

About myself. AI has made information way more accessible, opening up and facilitating brilliant ideas that once seemed beyond our reach. Yet, as LLMs and AI agents become part of our everyday lives, one question keeps floating through my mind: How much do we truly understand about the agents we increasingly rely on and the risks they encounter as they interact with the world? As we give them more powerful tools and greater freedom to act, is our ability to secure, understand, oversee, and control their behavior keeping pace? These questions stay with me, shaping both my curiosity and the research I choose to pursue.

Trustworthy AI Agents at the intersection of AI Security, AI Safety, and Evaluation & Trustworthiness

I study the cybersecurity threats and safety risks of AI systems, from prompt injection and privacy leakage to rogue agent behaviors and misalignment, so that progress does not come at the cost of security, safety, and meaningful human control.

Beyond research. I’m passionate about photography, books, and music, and I almost never say no to a new adventure. You can call me Judy; my Vietnamese name is Dung, pronounced /zʊŋ/.

News

Aug 01, 2026 Our paper FLOUR: Mixing the Right Ingredients for LLM Unlearning in Latent Space was accepted at Findings of EMNLP 2026.
Jul 23, 2026 Our paper Indelible Backdoors was accepted at ECCV 2026 as a poster. This work was led by Evelyn, an undergraduate I mentored. Congrats, Evelyn! 🎉
Jun 08, 2026 Started working as a Research Intern at Microsoft Research, Redmond, WA, on proxy evaluation and post-training of cybersecurity agents.
Jun 07, 2026 Our paper MorphoMod: Visible Watermark Removal with Morphological Dilation received the 🏆 Best Paper Award at the CVPR 2026 SAFE Workshop!
Jan 30, 2026 Our paper LOGSAFE was accepted at ICCPS’26, France.
Nov 10, 2025 Our paper SUGAR was accepted at WACV’26. See you in AZ!
Jun 21, 2025 One paper was accepted at ESORICS 2025.
May 27, 2025 Started my summer internship with Center for AI Safety CAIS
Apr 14, 2025 One paper was accepted at FSE2025 - IVR Track.
Apr 01, 2025 Presented our paper at HotSoS 2025.
Mar 26, 2025 Our paper PARDON: Privacy-Aware and Robust Federated Domain Generalization was accepted at ICDCS’25 (Scotland, UK).
Feb 27, 2025 Presented PBP at NDSS’25, San Diego, CA.
Jan 01, 2025 Successfully passed my Preliminary Exam.
Nov 21, 2024 My first PhD paper PBP was accepted at NDSS’25.
Dec 13, 2023 Presented my paper IBA at NeurIPS’23, New Orleans.
Aug 15, 2023 I started my first-year PhD @ Vanderbilt University.

Selected Projects

  1. infuse.png
    INFUSE: A Unified Surface Benchmark for Indirect Prompt Injection in Agentic Systems
    Thuy Dung Nguyen, Jingyi Guo , Rana Seda Dubauskas , and 3 more authors
    arXiv preprint arXiv:2610.20776, Oct 2026
  2. sugar.png
    SUGAR: A Sweeter Spot for Generative Unlearning of Many Identities
    Dung Thuy Nguyen, Quang Nguyen , Preston K. Robinette , and 3 more authors
    In 2026 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) , Mar 2026
  3. pbp.png
    PBP: Post-training Backdoor Purification for Malware Classifiers
    Dung Thuy Nguyen, Ngoc N Tran , Taylor T Johnson , and 1 more author
    Network and Distributed System Security Symposium (NDSS), 2025
  4. backdoor-survey.png
    Backdoor attacks and defenses in federated learning: Survey, challenges and future research directions
    Thuy Dung Nguyen, Tuan Nguyen , Phi Le Nguyen , and 3 more authors
    Engineering Applications of Artificial Intelligence, 2024
  5. iba.png
    Iba: Towards irreversible backdoor attacks in federated learning
    Thuy Dung Nguyen, Tuan A Nguyen , Anh Tran , and 2 more authors
    Advances in Neural Information Processing Systems, 2024
  6. fedgrad.png
    FedGrad: Mitigating Backdoor Attacks in Federated Learning Through Local Ultimate Gradients Inspection
    Thuy Dung Nguyen, Anh Duy Nguyen , Thanh-Hung Nguyen , and 4 more authors
    In 2023 International Joint Conference on Neural Networks (IJCNN) , 2023
  7. jolot.png
    Joint optimization of charging location and time for network lifetime extension in WRSNs
    Thuy Dung Nguyen, Tuyen Nguyen , Thanh Hung Nguyen , and 2 more authors
    IEEE Transactions on Green Communications and Networking, 2021