AI-powered record management for transitional justice: advancing transparency and accountability in post-conflict societies

This study examines the transformative potential and ethical risks of Artificial Intelligence (AI)-powered Electronic Records Management (ERM) systems in advancing transparency and accountability within transitional justice processes in post-conflict societies. Drawing on a qualitative comparative case study of Rwanda, Sierra Leone, and South Africa, the research investigates how AI technologies, including natural language processing, automated classification, and intelligent declassification, can enhance the accessibility, authenticity, and usability of truth commission archives and judicial records. Findings reveal four key opportunities: efficient processing of unstructured testimonial data, transparent declassification of sensitive materials, victim-centered reparations through automated claim validation, and preservation of collective memory via multilingual digital curation. However, these benefits are counterbalanced by significant context-dependent risks: algorithmic bias reinforcing dominant political narratives (particularly in Rwanda), digital exclusion marginalizing rural populations (especially in Sierra Leone), and surveillance weaponization threatening survivor safety. The study proposes a rights-based framework grounded in six principles, transparency, privacy, inclusivity, accountability, ethical design, and sustainability, to guide responsible AI deployment. This framework positions technology not as a neutral solution but as a socially embedded tool requiring participatory governance, independent oversight, and alignment with human rights norms. The research contributes empirical insights for policymakers, archivists, and peacebuilders seeking to harness AI for justice while safeguarding against its potential to reproduce inequalities in fragile contexts.