Automating safety: developing better data models to help foster prosocial platforms

This project identified and investigated how misunderstandings of harm and safety flow into flawed data logics and ineffective automated digital platform responses. A key output from the project included the journal article “Safety for Whom? Investigating How Platforms Frame and Perform Safety and Harm Interventions” (Gillett, Stardust, and Burgess 2023), which aimed to understand how five major digital platforms frame and define the issues of harm and safety, and identified the interventions they publicly report introducing to address these issues. Findings from this research importantly show that, in the absence of more meaningful measures that may foster safer digital cultures, at the heart of platform commitments is the scaling up of automated tools to moderate the enormous volume of content large digital platforms host.

Another key output included the journal article “Surveillance does not equal safety: Police, data and consent on dating apps” (Stardust, Gillett, and Albury, 2022). The article interrogated how dating apps focus on increasing user surveillance and other mechanisms that are deployed in the name of ‘safety’, rather than addressing the more complex questions about how their users experience and understand harm and safety, and how these understandings might better inform responses that enable users to feel safe. This paper was an extension of the article that the authors wrote for the Conversation (Gillett, Albury, Stardust, 2021).