Who Verifies the Verifiers? AI, Local Knowledge Production, and Credibility in Tanzania’s 2025 Election

Abstract

This study examines how Factchecking AI tools and community-based credibility practices interacted during Tanzania’s 2025 presidential election. Using data from five regions, we analyzed 342 media materials, 87 interviews, field observations, and 22 distinct contested election events. The findings identify three forms of verification disjuncture. First, AI-assisted translation and transcription sometimes changed the meaning of politically important statements before journalists could verify them. Second, fact-checks and disputed claims often moved through different communication channels, which limited the chances that people would see the correction. Third, some communities trusted personal experience, familiar sources, and collective discussion more than professional fact-checkers. Participants judged credibility through relationships, lived experience, group discussion, humor and satire, and social accountability. These practices sometimes supported the same conclusions as institutional verification and sometimes produced different judgments. We argue that fact-checking becomes politically effective when the meaning of a claim is preserved, the correction reaches the people discussing it, and the source of the correction is recognized as credible. This explains why an accurate fact-check can still have limited public influence.