Generative artificial intelligence has transformed the production of political disinformation, lowering the cost of fabricating convincing synthetic audio, video, and text to nearly zero and enabling forms of automated, personalised manipulation that older propaganda techniques could not sustain. This article offers a critical review of what this shift means for electoral security in Nigeria, a democracy already marked by ethno-religious fragmentation, a winner-takes-all political culture, and uneven trust in electoral institutions. Drawing on conceptual and theoretical literature, empirical evidence from Nigeria’s 2023 general election, and an assessment of the regulatory and policy landscape, the review argues that AI functions less as a novel political actor than as a force multiplier for disinformation practices that predate it, gaining its electoral significance from how it interacts with, rather than replaces, existing Nigerian vulnerabilities. Documented incidents from the 2023 election, including deepfake endorsements, fabricated audio alleging electoral rigging, and the liar’s dividend invoked around a genuine campaign gaffe, illustrate how synthetic content exploited institutional weaknesses such as the technical failures of the Bimodal Voter Accreditation System and the IReV results portal. The review finds Nigeria’s regulatory response reactive and fragmented, constrained by definitional gaps, overlapping institutional mandates, weak technical verification capacity, and an unresolved tension between restricting harmful content and protecting legitimate political speech. It concludes that closing this gap requires Nigeria-specific empirical research on actual voter response, longitudinal and comparative African evidence, and interdisciplinary work connecting technological, institutional, and legal capacity, since no single one of these alone accounts for how AI-enabled disinformation threatens electoral security.