L'IA générative dans les campagnes électorales : cartographie des tendances mondiales

The report finds that 80% of countries with competitive national elections in 2024 experienced GenAI-related incidents, with 69% of these incidents having a harmful impact. The majority of GenAI use focused on content creation (90%), and nearly half of the incidents (46%) had unknown sources. While some positive uses for civic engagement were noted, the document emphasizes the risks of disinformation and foreign interference, calling for mandatory disclosure standards and improved detection mechanisms.

• Generative AI was prevalent in the majority of competitive national elections in 2024, with 80% of the 50 analyzed countries experiencing GenAI-related incidents. India and the United States recorded the highest number of incidents, with 30 each.
• The primary application of GenAI in elections is the production of synthetic content, including audio, images, videos, and social media posts, which accounted for 90% of recorded incidents. These tools are often integrated into broader influence operations using personalized messaging and targeted distribution.
• Attributing GenAI content to a specific source is difficult, as 46% of incidents had unknown origins. Among identifiable sources, candidates and political parties were responsible for 25%, foreign actors for 20%, and other groups (such as lobbyists or individuals) for 9%.
• Most GenAI applications in elections were harmful, with 69% of incidents categorized as having a detrimental role. Notably, 100% of incidents attributed to foreign actors were found to be harmful, such as the coordinated use of AI-generated voice content by Russian and Chinese actors during the Solomon Islands elections.
• While predominantly harmful, GenAI has been used constructively for civic engagement and accessibility. For example, 38% of uses by national political parties and candidates were beneficial, including the use of AI chatbots in Japan to answer voter questions and encourage turnout.
• The report highlights specific cases where GenAI significantly impacted electoral outcomes or perceptions, such as in Romania, where coordinated AI manipulation contributed to the annulment of presidential election results, and in Bangladesh, where a deepfake video falsely announced a candidate's withdrawal.

APA

International Panel on the Information Environment (2025). L'IA générative dans les campagnes électorales : cartographie des tendances mondiales. https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/690a0c8b6525d26d8c8c7906_Final%20FR%20Elections%20SFP.pdf

Chicago

International Panel on the Information Environment. L'IA générative dans les campagnes électorales : cartographie des tendances mondiales. 2025. https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/690a0c8b6525d26d8c8c7906_Final%20FR%20Elections%20SFP.pdf.

Wikipedia

{{cite report |author=International Panel on the Information Environment |title=L'IA générative dans les campagnes électorales : cartographie des tendances mondiales |date=2025 |url=https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/690a0c8b6525d26d8c8c7906_Final%20FR%20Elections%20SFP.pdf |access-date=23 October 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{internationalpanelontheinformationenvironment2025lia, author = {{International Panel on the Information Environment}}, title = {{L'IA générative dans les campagnes électorales : cartographie des tendances mondiales}}, institution = {International Panel on the Information Environment}, year = {2025}, url = {https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/690a0c8b6525d26d8c8c7906_Final%20FR%20Elections%20SFP.pdf}, urldate = {2026-10-23}, note = {Indexed by Climate Insights Directory} }