Modeling False Claims Against Politicians Alleged on Social Media

Prof. Stéphane Gagnon’s work examines disinformation, misinformation, and information manipulation in the political arena, with a particular focus on corruption allegations against politicians. His research looks at both false allegations and legitimate accusations, aiming to clarify them as quickly as possible so that valid cases can proceed appropriately while false claims can be identified and addressed. Rather than focusing mainly on what circulates on social media, his work studies how democratic institutions such as parliaments, congresses, and senates respond to allegations through debates, political party reactions, and mitigation practices.

A central contribution of his research is its focus on parliamentary institutions and the legislative branch, which he argues have been neglected in both public administration and cybersecurity discussions. His work identifies key audiences who can benefit from this research, including political staff, media specialists, fake-news debunkers, parliamentary administrators, cybersecurity professionals, and technical researchers developing datasets and AI tools. Through case studies, workshops, surveys, and collaboration with practitioners, his research explores how institutions can prepare for growing volumes of disinformation, including future threats such as deepfakes.

Prof. Gagnon’s recent work also develops technical approaches for analyzing and responding to disinformation. His team has built datasets and case studies, used the DISARM Foundation framework to study disinformation campaigns, and explored large language models combined with knowledge graphs through GraphRAG. He is also advancing the idea of “debate graphs” to support plurality, freedom of speech, and institutional control over disambiguation processes, rather than relying entirely on commercial AI systems. His broader goal is to help democratic institutions build durable, transparent, and bias-aware tools for understanding political debates, assessing allegations, and combating disinformation without enabling censorship.

To learn more about the tool developped, please visit the MAIDGAR site.