Understanding the role of contexts in managing privacy online through AI-agents.

In online social systems, users are allowed to create and share content about themselves and others. The content being shared can affect anybody involved in it, and the privacy of individuals can be compromised. Existing work has shown that the use of artificial intelligence (AI) techniques can reveal even more information about the content such as faces, objects or location. This calls for the development of new privacy-enhancing technologies (PETs) to protect individuals online.

In this project, it is crucial to understand privacy preferences of the users that can be different in various contexts. One formulation of privacy is contextual integrity, which understands privacy in a social context. Context depends on many factors such as location, people, activity and such. Contexts are useful to decide if it is appropriate for an individual to share information. For example, it would be appropriate for an individual to share their medical history with doctors in a medical context but not in a context where the user is trying to access a particular web site. To this extent, we would like to understand the role of contexts in privacy decisions better.