Kai Yao

My name is Kai Yao, and I am currently a PhD candidate in the School of Informatics at the University of Edinburgh, under the mentorship of Dr. Marc Juarez. My research focuses on secure and trustworthy machine learning, encompassing areas such as security, privacy, and fairness in artificial intelligence systems.

I began my academic journey at Fudan University, where I earned my Bachelor’s degree. I then pursued a Master’s degree at Johns Hopkins University, concentrating on AI applications in bio-science. Professionally, I have contributed as an AI frameworks engineer at Intel and as an AI algorithm engineer at Huawei. Beyond my professional endeavors, I am passionate about exploring the ethical dimensions of AI and advocating for responsible technology development in this rapidly evolving field.

News

• New work published: AuthPrint: Fingerprinting Generative Models Against Malicious Model Providers. AuthPrint introduces a fingerprinting method to detect when a black-box image-generation API quietly swaps the claimed model for a lower-quality one—helping protect users from malicious providers.

• Talk @ IEEE SaTML 2025 (Copenhagen): I presented our work SoK: What Makes Private Learning Unfair?. See the short recap on Bluesky and event details at SaTML 2025.

• Paper acceptance: SoK: What Makes Private Learning Unfair? was accepted to the 3rd IEEE Conference on Secure and Trustworthy Machine Learning (SaTML 2025). A preprint is available on arXiv.