[CS] Zhuolin Yang Dissertation Defense/Apr 16, 2025

[via cs]

This is an announcement of Zhuolin Yang's Dissertation Defense.
===============================================
Candidate: Zhuolin Yang

Date: Wednesday, April 16, 2025

Time: 11:30 am CST

Remote Location: https://uchicago.zoom.us/j/93431374062?pwd=nzPDvhYKg4XJfDETUuCylrcKAtUETt.1 (Meeting ID: 934 3137 4062 & Passcode: 330185)

Location: JCL 346

Title: Identifying and Mitigating AI-enhanced Privacy Attacks

Abstract: In the security community, there is a long history of identifying privacy risks and attacks from exposure of data to third parties. Over time, attacks have become more powerful and effective at extracting private data from ordinary observations. For much of this century,  most privacy attacks on various data modalities and computing platforms have grown to rely on techniques derived from statistical machine learning. But this landscape is changing, thanks to the arrival of significantly different AI models and architectures, capable of extracting patterns and information once considered beyond the reach of ML models. User data once thought to be “squeezed dry” is now revealing entirely new and unexpected results through these advanced AI techniques. This, in turn, requires us to reevaluate and redefine what is possible in the field of privacy attacks. 

In this defense, I will present my research in this direction, focusing on reevaluating privacy attacks in personal computing by integrating and adapting advanced AI/ML algorithms. Specifically, I will discuss my work on adapting these methods to fundamentally redesign two well-known privacy attacks, keystroke inference and UI confusion/hijacking, significantly amplifying their impact on users while evading detection. These findings highlight the urgent threat posed by AI-enhanced attacks and emphasize the need to study and develop more robust defenses. 

Advisors: Heather Zheng and Ben Zhao

Committee Members: Ben Zhao, Heather Zheng, Pedro Lopes, and Grant Ho