The use of AI-generated images in the Colombian Electoral Cycle 2026

Sammanfattning

This thesis examines how generative artificial intelligence was used in political campaigns on platform X during the 2026 Colombian elections. It conducted a Qualitative Content Analysis (Saldaña, 2013) of fifty AI-generated visual artifacts collected after the March 8 parliamentary elections. The findings reveal three main strategies: visual stigmatization, emotional mobilization, and heroic fabrication. To understand why these strategies are effective, the research draws on Han's (2022) concept of infocracy and BehnamShad's (2025) theory of algorithmic hegemony. These frameworks help explain the structural conditions that make these approaches rational for campaign producers. At the micro level, it relies on research related to multimodal political communication and populist visual propaganda. (Adebomi, 2025; Hepp et al., 2023; Rodarte et al., 2023) to analyze the specific features of the visual content. In the 2026 Colombian elections, AI-generated visuals served as powerful tools for political polarization. Rather than relying on argument, these images flooded the digital space with emotionally charged content that was easy to share and inexpensive to produce. While much of the literature focuses on the liar's dividend (Schiff et al., 2025), research shows that campaigns in Colombia proactively used AI to build persuasive political narratives. This strategy was shaped by high levels of institutional distrust, a sophisticated system for paid amplification (Bodegas Digitales), and a history of using politically sensitive imagery. Mapping these factors highlights a specific context for the use of generative AI and suggests that similar environments may exhibit comparable patterns. The thesis offers empirical evidence of AI use in a setting increasingly important to society and politics, and argues that focusing solely on defensive uses of AI-generated media overlooks AI's ability to shape political realities.