Date: September 25, 2026
Speaker: Eugene Bagdasarian, Assistant Professor, UMass Amherst
Title: AirGapAgent: Protecting Privacy-conscious AI Agents
Abstract: The growing use of large language model (LLM)-based agents to manage sensitive user data raises significant privacy concerns. While these agents excel at understanding and operating across contexts, this capability can be exploited by adversaries. We introduce a novel threat model where a malicious third-party manipulates the context of interaction to trick AI agents into revealing private information not relevant to the task at hand.
Grounded in the framework of contextual integrity, we introduce AirGapAgent, a privacy-conscious agent designed to prevent unintended data leakage by restricting the agent's access to only the data necessary for a specific task. Extensive experiments using Gemini, GPT, and Mistral models as agents validate our approach's effectiveness in mitigating this form of context hijacking while maintaining core agent functionality. For example, we show that a single-query context hijacking attack on a baseline Gemini-based agent reduces its ability to protect user data from 94% to 45%, while an AirGapAgent achieves 97% protection, rendering the same attack ineffective. Finally, we discuss future directions including contextual security and designs for agentic policy engines.
Bio: Eugene Bagdasarian is an Assistant Professor at University of Massachusetts Amherst and a Researcher at Google. His work focuses on studying attack vectors in AI systems deployed in real life and proposing new designs that mitigate these attacks. Previously, he received the Distinguished Paper Award at USENIX Security and Apple AI/ML Ph.D. Fellowship.