| This study investigates the effects of AI‑generated political videos with past‑oriented versus future‑oriented narratives on message evaluation and subsequent political participation intentions among adults in Bangladesh. In an online experiment using a mixed factorial design, participants (n = 192) were randomly assigned to view one of two short campaign‑style AI‑generated videos: a nostalgia clip emphasizing recent democratic struggle and a visionary clip highlighting future reforms and long‑term civic renewal. Pre‑ and post‑exposure scales measured attitudes toward AI‑generated content and AI‑driven political misinformation, while post‑exposure measures assessed perceived message effect, narrative involvement, perceived credibility and realism, trust in AI as a political information source, political participation intentions, and voting likelihood. The results showed that the future‑oriented video (visionary) overall performed better on key outcomes, producing higher perceived message effect, greater narrative involvement, stronger political participation intentions, more favorable overall evaluations, and a higher likelihood of voting than the past‑oriented video (nostalgia). Perceived credibility and post‑exposure attitudes toward AI‑generated content also trended in the same direction. Exposure to either AI‑generated video led to significant increases in both acceptance of AI‑generated content and concern about AI‑driven political misinformation, indicating that participants became simultaneously more open to and wary of AI in politics. Demographic and media‑use analyses demonstrated that age, education, and gender were associated with differences in misinformation concerns, message evaluations, and voting likelihood, but did not consistently predict participation intentions. Multiple regression analyses identified perceived credibility and realism of the video and trust in AI as a political information source as the strongest positive predictors of political participation intentions, while pre‑exposure misinformation concern was a negative predictor and short‑video viewing frequency had a small positive effect. Keywords: AI-generated political video, narrative engagement, message effectiveness, political participation intentions, AI-trust |