Privacy Agents: Utilizing Large Language Models to Safeguard Contextual Integrity in Elderly Care

Abstract

A value-based design process in the development of robotic technologies for elderly care requires approaches to protect privacy. With the rise of Large Language Models (LLMs) new use cases for robotic technology can be facilitated. In this paper we present a conceptual approach to utilizing LLMs to enhance privacy. We refer to a use case of a care documentation support agent that should aid care workers in their care routines. Our contribution is based on the understanding of Nissenbaum’s privacy as contextual integrity. We introduce a privacy agent that continuously monitors information flows of recorded conversations, and identifies key parameters that are compared to the contextual privacy norms to detect privacy violations.