“Friends” of Science? Critically analysing the multimodal discourse of a long-standing climate denial front group on Instagram

Problematic climate information affects public support for mitigation policies. Frequently distributed through a powerful coalition of contrarian actors—the climate change counter movement (CCCM)—it undermines scientific evidence to protect vested interests through ongoing promotional campaigns. On social media, image-based climate misinformation often outperforms verified content. Instagram, with its multimodal vernacular, is a crucial platform for climate misinformation research. In this paper, a novel, visual-first exploratory method combines unsupervised machine learning with multimodal critical discourse analysis. I employ a custom computational tool, the ‘Image Machine, to identify dominant visual signatures and discursive themes in a dataset of 15,000 Instagram posts, collected via hashtags representing climate denial discourse.Cluster analysis reveals a distinct visual signature that represents the branding logics of established contrarian front group, Friends of Science Society (FoS). Through multimodal critical discourse analysis of a visual cluster and representative post case study, this research interrogates the discursive strategies and climate denial rhetoric of FoS that—through deliberate intertextual, multimodal tactics that exploit Instagram’s affordances—aim to confound public climate science knowledge. Findings situate FoS firmly in the role of front group within the denial machine, instrumentally using climate denial as a rhetorical and discursive strategy to support Canada’s fossil fuel interests.Funded by fossil interests, FoS exploits signature visuality to solidify contrarian narratives and reinforce its brand. In documenting individuals and organisations implicit in the FoS funding structure, this research contributes to scholarship about the CCCM, as well introduces a novel methodology to ascribe multimodal—particularly, visual—means for detecting problematic climate information.