Conspiracy Theories and Political Polarization: A Study of Brazilian Far-Right Narratives on Telegram

This study analyzes how conspiracy theories function as a strategy for political mobilization and the exercise of political violence within far-right groups in Brazil from 2020 to 2024. Based on real-time data collection from Telegram, the research examines a range of conspiratorial communities, including anti-vaccine and off-label medication groups that monetize false miracle cures, climate denialism and anti-science communities that spread misinformation, anti-woke and anti-gender groups that propagate hate speech and neo-Nazi ideologies, apocalyptic, survivalist, and esoteric groups that merge occult beliefs with conspiracy narratives, and anti-communist movements that serve as a core ideological framework. These theories are instrumental in fostering political polarization during critical moments, such as civic holidays, elections, and events resulting from climate crises, like the 2024 floods in the South of Brazil. We utilize the GPT-4 Omni API for automated detection of conspiracy theories and image clustering, alongside network analysis of text components. Instead of relying solely on traditional topic modelling, we extract grammatical components using Part-of-Speech tagging, followed by semantic labelling to capture key roles like object, subject, and action (verb). Verbs are then clustered using OpenAI's embedding model, enabling a nuanced analysis of language use within these conspiratorial groups. The research highlights the role of conspiracy theories in escalating violence and undermining democratic structures. This study emphasizes the need to understand the socio-cultural roots of these narratives, offering a sociological perspective on how they resonate within polarized populations and contribute to the ongoing erosion of trust in mainstream epistemic authorities.