Evaluating Automated System Interventions Against Email Harassment

Email and other forms of electronic communication are becoming increasingly more essential to our everyday lives. However, with this growth comes the paralleled increased risk of email harassment, exacerbated by the current lack of platform support for managing these harmful messages. This paper explores different interfaces for the automated detection and management of email harassment using artificial intelligence in order to investigate what degree of platform intervention email users prefer when navigating their email platform. Through conducting user studies involving three different email platform prototypes based on the Gmail platform, we employ mixed-method analysis to evaluate how varying levels of platform intervention affect users’ perceived sense of safety, agency, and trust with their email platform. Our primary findings suggest that users generally benefited from each of the system intervention strategies and desired higher intervention features when combating email harassment, as well as ways of managing this intervention based on their unique preferences.