Kai Yao

About

I am Kai Yao, a PhD candidate in the School of Informatics at the University of Edinburgh, working under the supervision of Dr. Marc Juarez. My research centers on secure and trustworthy machine learning, with emphasis on security, privacy, and fairness in AI systems. My current work examines provenance in image-generation models through techniques such as model watermarking and model fingerprinting.

I completed my Bachelor's degree at Fudan University and a Master's degree at Johns Hopkins University with a focus on AI for bioscience. I have also worked in the industry for 3 years as an AI algorithm engineer at Huawei and Intel, prior to joining Edinburgh.

News

2 0 2 5 . 1 2 Paper accepted: Smudged Fingerprints: A Systematic Evaluation of the Robustness of AI Image Fingerprints accepted to the 4th IEEE Conference on Secure and Trustworthy Machine Learning (SaTML 2026). Preprint on arXiv.
2 0 2 5 . 0 9 New work: AuthPrint: Fingerprinting Generative Models Against Malicious Model Providers. It detects silent model swapping by malicious model providers in black-box image-generation APIs.
2 0 2 5 . 0 4 Talk (SaTML 2025, Copenhagen): Presented SoK: What Makes Private Learning Unfair?. Short recap on Bluesky and program details at SaTML 2025.
2 0 2 4 . 1 2 Paper accepted: SoK: What Makes Private Learning Unfair? accepted to the 3rd IEEE Conference on Secure and Trustworthy Machine Learning (SaTML 2025). Preprint on arXiv.

Publications

Yao K, Juarez M.

Smudged Fingerprints: A Systematic Evaluation of the Robustness of AI Image Fingerprints

arXiv preprint, 2025. To appear in 2026 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML).
Yao K, Juarez M.

AuthPrint: Fingerprinting Generative Models Against Malicious Model Providers

arXiv preprint, 2025.
Yao K, Juarez M.

SoK: What Makes Private Learning Unfair?

2025 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML).
Rochman ND^*, Yao K^*, Perez-Gonzalez NA^*, Wirtz D, Sun SX.

Single-Cell Volume Measurement Utilizing the Fluorescence Exclusion Method (FXm)

Bio-protocol, 2020.
Yao K^*, Rochman ND^*, Sun SX.

CTRL: A Label-Free Artificial Intelligence Method for Dynamic Measurement of Single-Cell Volume

Journal of Cell Science, 2020.
Perez-Gonzalez NA^*, Rochman ND^*, Yao K^* et al.

YAP and TAZ Regulate Cell Volume

Journal of Cell Biology, 2019.
Yao K^*, Rochman ND^*, Sun SX.

Cell Type Classification and Unsupervised Morphological Phenotyping from Low-Resolution Images Using Deep Learning

Scientific Reports, Nature Publishing Group, 2019.
Zhang Q, Meng Z, Zhang Y, Yao K et al.

Phantom-Based Experimental Validation of Fast Virtual Deployment of Self-Expandable Stents for Cerebral Aneurysms

BioMedical Engineering OnLine, 2016.

Note: ^* denotes equal contribution (co-first author).

Teaching

Teaching Assistant

Privacy and Security with Machine Learning

University of Edinburgh

2023–2025
Mathematical Image Analysis

Johns Hopkins University

2019–2020

Research Tutoring & Mentoring

Johns Hopkins (UG): Felipe Takaesu · Eliana Crentsil · Lucia Sablich
Johns Hopkins (PG): Shannon Flanary
University of Edinburgh (PG): Chunhan Fang