The proliferation of false information poses significant challenges to democracy worldwide. Concerns grow about the effects of Artificial Intelligence (AI) misuse on political trust and democratic attitudes as emerging technologies like deepfakes, synthetic news, and AI-powered personalised political advertising reshape information ecosystems. Using a survey vignette experiment fielded in Australia and Canada, we randomly expose respondents to information about AI-related threats to elections, varying how such risks are acknowledged by different actors, and measure the effects on trust in electoral outcomes and support for democracy. The two countries offer a useful comparative setting given their differing electoral administration models, compulsory versus voluntary voting systems, and emerging approaches to AI regulation. The findings offer cross-national experimental evidence on how electoral management bodies, media, and policymakers might communicate about AI-related risks without eroding democratic trust, with direct implications for strengthening electoral integrity across contexts.
