Multimodal Misinformation: How TikTok’s Visual Content Shapes Climate Obstruction Claims

Climate misinformation continues to evolve, employing strategies that delay climate action. These range from outright denial to subtler forms of disbelief that acknowledge some science while resisting its implications, solutions, or urgency (Ekberg et al., 2023; Coan et al., 2021). While social media platforms have been shown to accelerate the diffusion of such claims (Treen et al., 2020; Galaz et al., 2023), much of the existing research focuses on text-based platforms like Twitter or Facebook (Gounaridis & Newell, 2024; Böhm & Pfister, 2024; Bloomfield & Tillery, 2019), often neglecting the multimodal affordances of video-centric platforms like TikTok (Baltasar et al., 2024; Sun et al., 2024). These affordances— combining visuals, audio, text overlays, and soundtracks—create expressive narratives and present distinct challenges for detecting misinformation. Despite its growing influence among Gen-Z users and its critical role in shaping climate attitudes and discourse, the platform remains significantly underexplored in this context. This study investigates the multimodal and expressive nature of climate misinformation on TikTok, focusing on content under the hashtags #weathermanipulation and #weathercontrol. These hashtags are associated with extreme weather events, acute manifestations of climate change, and narratives of climate obstruction (Baltasar et al., 2024; CAAD, 2024). While previous studies suggest TikTok has less climate misinformation (Basch et al., 2022) also compared to platforms like YouTube or Instagram (Duan et al., 2024), content within these clusters demonstrates how misinformation adapts to TikTok’s unique expressive affordances. Using a multimodal content analysis approach, this study aims to examine the interplay of visuals, text overlays, narration, and music in shaping misinformation narratives. Key research questions include: What themes emerge in #weathermanipulation and #weathercontrol content? How do these narratives, and their multimodal and expressive elements, evolve in response to extreme weather events? To address these questions, the study analyzes publicly available TikTok posts with these hashtags, collected via automated scripting. The dataset includes videos, engagement metrics, and post dates. Methodologically, we develop a computational multimodal framework analyzing visuals, music sentiment, and text overlays and transcripts. These embeddings are integrated into a multimodal topic modeling framework (Multimodal BERTopic) to identify emergent themes and their prevalence over time, followed by qualitative analysis for thick descriptions and deeper insights. This study makes three key contributions. Empirically, it identifies novel forms of climate misinformation. Methodologically, it demonstrates the feasibility of computational multimodal analysis. Theoretically, it expands the understanding of climate claims to include multimodal dimensions. By addressing these gaps, the study offers critical insights into the evolving nature of climate misinformation on TikTok, highlighting the platform’s unique role in the misinformation ecosystem, where content often resists traditional text-based detection methods.