In this work, we present a practical design framework for a privacy-aware virtual assistant for everyday AR, focusing on educating users with a lack of knowledge regarding technical and/or privacy literacy. Our approach features human-in-the-loop to learn privacy context, provides transparency into the system state of privacy detectors, and affords the user control and the ability to provide feedback to the system. The target interaction allows users to correct and customize the privacy-aware assistant’s decision on the information it considers private to create a more personalized and user-driven approach to privacy management. Additionally, this framework affords inclusivity, offering a straightforward interface to explain machine decisions and educate users about privacy implications, making privacy settings accessible to users of diverse backgrounds, cultural origins, and abilities. This research looks to contribute to the IEEE VR and IDEATExR research communities to establish ethical practices regarding privacy in the future of everyday and pervasive AR.
This work is currently in progress.