Privacy-Enhancing Technologies for AI Systems: A Tutorial

Open Webpage

Liv Galliker d'Aliberti, Evan Gronberg, Joseph Kovba

Published: 21 Jun 2024, Last Modified: 06 Nov 2025CrossrefEveryoneRevisionsCC BY-SA 4.0
Abstract: This tutorial presents privacy threats to artificially intelligent (AI) systems and proposes the use of several privacy-enhancing technologies (PETs) to address them. Such threats can affect both model owners and system users, be internal or external to the system, and manifest at multiple stages of the AI lifecycle (i.e., the model development, deployment, and inference phases). PETs are next-generation technologies, uniquely addressing threats to "data-in-use" processes that have previously gone unprotected. In this tutorial, we explore the impact of PETs on a variety system-level characteristics.