AI governance, disinformation, and democratic resilience: mapping the 2019–2025 literature and developing a Taiwan-informed framework

Artificial intelligence (AI) can intensify disinformation and support verification, creating a governance problem of institutional coupling: how distributed civic and technical signals translate into democratic response. This study combines a reconstructed 2019–2025 literature corpus (N = 211) with a source-verified Taiwan evidence synthesis. Bibliometric mapping identifies intersections among AI, disinformation, media, deepfakes, democracy, governance, platforms, and trust. Taiwan evidence most strongly supports sensing and verification, partially supports translation/escalation and response/accountability, and does not directly establish longitudinal adaptation. We therefore develop a Taiwan-informed bottleneck framework in which democratic resilience depends on verification crossing translation, authority/accountability, and learning interfaces.