Exploring opportunities, risks, and resilience among young indigenous social media users in the Chittagong hill tracts through qualitative and quantitative lenses

Abstract

This study explores the digital experiences of Indigenous youth in the Chittagong Hill Tracts (CHT) of Bangladesh by combining qualitative inquiry with machine learning analysis. Through interviews and focus group discussions, we uncover how Indigenous social media users navigate challenges such as cultural misrepresentation, online harassment, censorship, and surveillance. Participants shared how these threats often extend beyond the digital realm into their everyday lives, and how they adopt strategies like social ciphering and selective visibility to resist and survive in online spaces.

To complement these narratives, we developed a multimodal machine learning model trained on the Fakeddit dataset to detect various forms of fake content, including manipulated, satirical, and imposter content. While the model demonstrated high accuracy in detecting manipulated content, closely aligning with the types of harmful posts reported by participants, it struggled with subtle misinformation categories, exposing key limitations of general-purpose moderation systems.

By integrating insights from Critical Race Theory and Indigenous HCI, we highlight the mismatch between algorithmic moderation and culturally situated online harm. Our findings inform the design of participatory, context-aware interventions that center Indigenous voices in the development of content moderation technologies.