Air Gap: Protecting Privacy-Conscious Conversational Agents

arXiv:2405.05175v1 Announce Type: new
Abstract: The growing use of large language model (LLM)-based conversational agents to manage sensitive user data raises significant privacy concerns. While these agents excel at understanding and acting on context, this capability can be exploited by malicious actors. We introduce a novel threat model where adversarial third-party apps manipulate the context of interaction to trick LLM-based agents into revealing private information not relevant to the task at hand.
Grounded in the framework of contextual integrity, we …

adversarial agents air apps arxiv can context cs.cl cs.cr cs.lg data excel exploited gap language large large language model llm malicious malicious actors manage novel party privacy privacy concerns protecting sensitive third third-party threat threat model understanding user data