๐Ÿ”ฌ Research

Trustworthy AI, with a focus on privacy, security, and robustness in large language models (LLMs).

๐Ÿ›ก๏ธ My work explores:
- ๐Ÿ”’ Adversarial robustness
- ๐Ÿ” Differential privacy
- ๐Ÿ’ง Watermarking techniques

to ensure reliable and ethical deployment of LLMs in real-world applications.

๐Ÿ“š Publications

๐Ÿ“– Published

    Kieu Dang, Phung Lai, NhatHai Phan, Yelong Shen, Ruoming Jin, Abdallah Khreishah. ๐›ฟ-Steal: LLM Stealing Attack with LDP. In Proceedings of Asian Conference of Machine Learning (ACML 2025).

    Dylan Tarace, Phung Lai, Kieu Dang, Unal Tatar. AI-Powered Assessment of Wazuh for Obfuscated Threat Detection. In Proceedings of IEEE Systems and Information Engineering Design Symposium (SIEDS 2025).

    Kieu Dang, Phung Lai. Navigating Trustworthiness in LLMs: An Examination of Privacy, Security, and Robustness. In Proceedings of Computational Data and Social Networks (CSoNet 2024).

๐Ÿ“ Forthcoming (Submitted)

    Kieu Dang, Phung Lai, NhatHai Phan, Yelong Shen, Ruoming Jin. Watermark Paper. ACM CCS 2025.

    Kieu Dang, Phung Lai, NhatHai Phan, Yelong Shen, Ruoming Jin, Abdallah Khreishah, My Thai. SoK: Are Watermarks in LLMs Ready for Deployment? IEEE S&P 2025.

๐Ÿ“œ Patent

    Kieu Dang, Phung Lai, NhatHai Phan. Watermark Technique โ€” Filed non-provisional US patent, April 15, 2025.

๐Ÿ›๏ธ Research Experience

  • Responsible AI Lab โ€“ State University of New York at Albany Jan 2024 โ€“ Present
    ๐Ÿค Collaborating with Microsoft, NJIT, KSU, UF, and HBKU (Qatar) on trustworthy machine learning experiments.
    ๐Ÿงช Designing and evaluating adversarial robustness strategies for LLMs.