The Privacy Paradox of GenAI: What It Knows, Why It Matters, and How We Fight Back

Roberto Yus, 12-1pm Wed. Nov 5, 2025, online

Generative AI (GenAI) systems like ChatGPT, Claude, and Gemini are transforming how we work, communicate, and create, but they also pose serious privacy concerns. These systems often absorb sensitive inputs, operate atop opaque training data, and may retain more than users realize. The real paradox is this: while GenAI can compromise our privacy, it can also be harnessed to help protect it.

This talk explores both sides of that paradox. We begin by examining what these systems "know", from how everyday prompts may reveal personal or proprietary information to how models process, store, or share data. We then discuss why these risks matter, both for individuals and institutions navigating a rapidly shifting AI landscape. But this is not just a critique, it's a call to action. We'll share practical recommendations for using GenAI tools more safely and privately. Then, we'll turn to new research from the DAMS Lab that flips the script: using GenAI to interpret and explain complex corporate privacy policies, empowering users to make informed, privacy-conscious decisions. This talk is for anyone who wants to benefit from GenAI without surrendering control over their data, and who sees privacy not as a casualty of AI progress, but as a challenge AI itself can help us solve.

Roberto Yus is an Assistant Professor in UMBC's CSEE Department, where he leads the DAMS (DAta Management & Semantics) research group. His research interests concern Data Management, Knowledge Representation, the Internet of Things, and Privacy. Specifically semantic and privacy preserving data management in IoT environments.

Online