Transparent Moderation: Designing Explainable AI Interfaces for User Appeals in Content Moderation

While papers like Gao et al. (2025) highlight the opacity of current generative AI moderation systems and Divon et al. (2025) expose the epistemic injustice caused by ambiguous, gaslighting communications, most platforms still lack transparent, user-friendly explanations for moderation actions. This research proposes creating explainable AI interfaces that not only show users why content was moderated (with accessible, plain language justifications), but also guide users through a structured, supportive appeals process. The system could leverage state-of-the-art LLMs but with an added “explanation module” trained on real user queries and feedback. By focusing on marginalized communities—those most affected by moderation errors (Flynn et al., 2025; Lyu et al., 2024)—the project would iteratively co-design these interfaces, ensuring accessibility and trust. The novelty here lies in combining explainability (from AI/ML literature) with participatory UX design and policy transparency, moving beyond “accept/reject” buttons to a dialogic, restorative approach. This could significantly improve user trust, reduce perceptions of platform gaslighting, and set a new standard in digital rights.

References:

1. Platform gaslighting: A user-centric insight into social media corporate communications of content moderation. Tom Divon, Carolina Are, Pam Briggs (2025). Platforms & Society.
2. "I Cannot Write This Because It Violates Our Content Policy": Understanding Content Moderation Policies and User Experiences in Generative AI Products. Lan Gao, Oscar Chen, Rachel Lee, Nick Feamster, Chenhao Tan, M. Chetty (2025). arXiv.org.
3. Content Moderation and Community Standards: The Disconnect Between Policy and User Experiences Reporting Harmful and Offensive Content on Social Media. Asher Flynn, Z. Vakhitova, Lisa J. Wheildon, Bridget Harris, Brady Robards (2025). Policy & Internet.
4. "I Got Flagged for Supposed Bullying, Even Though It Was in Response to Someone Harassing Me About My Disability.": A Study of Blind TikTokers’ Content Moderation Experiences. Yao Lyu, Jie Cai, Anisa Callis, Kelley Cotter, John M. Carroll (2024). International Conference on Human Factors in Computing Systems.