Current - Issue

Generative artificial intelligence (AI) has introduced a new and structurally distinct class of threat to electoral integrity in Nigeria: the capacity to produce hyper-realistic synthetic media — including voice-cloned audio, deepfake video, and AI-generated text — in Nigeria’s three dominant regional languages (Hausa, Yoruba, and Igbo) and in Nigerian Pidgin, at negligible cost and without technical expertise. Nigeria’s 2023 general election constituted a watershed moment in this regard, marked by a documented proliferation of deepfake content falsely implicating presidential candidates in ballot manipulation, ethno-religious sedition, and criminal conduct. As Nigeria approaches the 2027 general elections, the threat has escalated in both sophistication and scale. This paper investigates the mechanisms, vectors, and political economy of AI-mediated electoral manipulation in Nigeria’s multi-ethnic digital space, with particular attention to the intersection of generative AI and ethno-regional political identity. Drawing on a mixed-methods study comprising content analysis of 1,240 synthetic media artefacts identified during the 2023 election cycle, semi-structured interviews (n=48) with fact-checkers, platform moderators, civil society actors, and INEC officials, and a structured digital survey (n=720) administered across Nigeria’s six geopolitical zones, the paper develops a typology of AI-mediated electoral manipulation (AMEM) and proposes a four-pillar Democratic Consent Protection Framework (DCPF) for Nigeria. Findings reveal that 67.4% of surveyed registered voters encountered AI-generated political content in the 30 days preceding the 2023 election, that ethnic and religious targeting was the dominant manipulation modality (present in 71.3% of identified deepfake content), and that existing regulatory instruments — including the Electoral Act (2022), the Nigeria Data Protection Act (2023), and INEC’s nascent AI Division — are structurally inadequate to address the speed, scale, and linguistic specificity of AI-mediated electoral threats. The paper contributes the first empirically grounded typology of AI electoral manipulation in a Sub-Saharan African multi-ethnic democracy, and the first governance framework specifically designed for this context.