Deepfakes in the 2025 Canadian Election: Prevalence, Partisanship, and Platform Dynamics

Authors: Victor Livernoche, Andreea Musulan, Zachary Yang, Jean-François Godbout, Reihaneh Rabbany

Published: 2025-12-15 21:49:40+00:00

Comment: 4 pages, 6 figures

AI Summary

This study empirically analyzes the prevalence and circulation of AI-generated images (deepfakes) across X, Bluesky, and Reddit during the 2025 Canadian federal election. It found that 5.86% of election-related images were deepfakes, with right-leaning accounts sharing them more frequently, often with defamatory intent. Despite their presence, most deepfakes were benign and harmful ones drew minimal attention, suggesting a modest overall impact on political discourse.

Abstract

Concerns about AI-generated political content are growing, yet there is limited empirical evidence on how deepfakes actually appear and circulate across social platforms during major events in democratic countries. In this study, we present one of the first in-depth analyses of how these realistic synthetic media shape the political landscape online, focusing specifically on the 2025 Canadian federal election. By analyzing 187,778 posts from X, Bluesky, and Reddit with a high-accuracy detection framework trained on a diverse set of modern generative models, we find that 5.86% of election-related images were deepfakes. Right-leaning accounts shared them more frequently, with 8.66% of their posted images flagged compared to 4.42% for left-leaning users, often with defamatory or conspiratorial intent. Yet, most detected deepfakes were benign or non-political, and harmful ones drew little attention, accounting for only 0.12% of all views on X. Overall, deepfakes were present in the election conversation, but their reach was modest, and realistic fabricated images, although less common, drew higher engagement, highlighting growing concerns about their potential misuse.

Key findings

Approximately 5.86% of election-related images were deepfakes, with a higher prevalence among right-leaning accounts (9.24%) compared to left-leaning accounts (3.87%). Most detected deepfakes were benign or non-political, and harmful deepfakes garnered very little engagement (0.12% of total views), suggesting a modest overall impact on online political discourse. However, highly realistic fabricated images, though less common, received higher engagement, indicating a potential vulnerability.

Approach

The researchers analyzed 187,778 image-containing posts from X, Bluesky, and Reddit using a high-accuracy deepfake detection framework trained on diverse modern generative models. They further employed a vision-language model for communicative intent classification and a large language model to infer authors' political leanings, enabling a multi-layered analysis of deepfake prevalence, purpose, and impact.

Datasets

X, Bluesky, Reddit (187,778 posts from 2025 Canadian federal election), OpenFake dataset, DF40, GenImage

Model(s)

UNKNOWN

Author countries

Canada

← Previous