AI-Driven Disinformation, Deepfakes, And Electoral Integrity In Emerging Democracies

Abstract

The rapid diffusion of generative artificial intelligence has transformed electoral disinformation from conventional falsehoods into scalable forms of synthetic manipulation that challenge trust, accountability, and electoral legitimacy. This literature review examines how AI-driven disinformation, particularly deepfakes, synthetic media, AI-assisted political messaging, bot-supported amplification, and liar’s dividend dynamics, affects electoral integrity in emerging democracies. Drawing on 17 reviewed studies spanning experimental, computational, conceptual, policy, and mixed-method designs, the article synthesises the principal forms of AI-driven disinformation, the causal mechanisms through which they operate, and the governance responses proposed to address them. The review finds that the most consistent effects do not lie in direct vote conversion, but in authenticity confusion, epistemic instability, false recognition, declining trust in information environments, and delegitimation of electoral procedures and outcomes. Evidence on candidate evaluation and voting intention exists, but it is more mixed and conditional than evidence on distrust, uncertainty, and institutional vulnerability. The review also shows that platform regulation, detection systems, fact-checking, digital literacy, and institutional responses remain partial and uneven in their effectiveness, particularly where regulatory capacity and public trust are weak. Overall, the article argues that AI-driven disinformation threatens electoral integrity by degrading the epistemic and institutional foundations on which democratic legitimacy depends.