Abstract

The emergence of commercial generative artificial intelligence (AI) models used for visual fabrication and distributive AI algorithms that curate content on social media platforms has fundamentally transformed the contemporary news environment, creating unprecedentedly favorable conditions for the creation and propagation of visual disinformation. This thesis argues that the synergistic relationship between these two AI systems has democratized the means of fabricating intersubjective reality through visual representation – and even incentivizes it – with profound implications for journalism and functional political democracy. Drawing on an interdisciplinary theoretical framework, this thesis presents a case study of three stratified examples surrounding the 2024 US presidential election, tracing a trajectory across elite political actors, pseudo-journalists, and ordinary social media users, showing how each order of influence distinctively employed these AI systems synergistically, and how they were structurally incentivized to do so, to produce and distribute visual disinformation. Complementing this, a research-creation component – a fabricated political news report produced entirely using free, widely available generative AI tools – serves as a proof-of-concept demonstration of the unprecedented ease with which effective visual disinformation can now be created. Ultimately, this thesis suggests that these AI systems are cultivating an emergent regime of truth in which engagement metrics, rather than factual accuracy, increasingly determine ‘what counts as true’ for voters. In turn, news consumers veer toward either belief in nonfactual information or apathy toward factual information. Both outcomes corrode the intersubjective reality upon which functional political democracy depends, thus underscoring the need for stronger regulatory frameworks and reimagined platform algorithms.