The intersection of artificial intelligence and political campaigning took a bizarre turn recently when former President Donald Trump shared a series of AI-generated images on his Truth Social platform. These images depicted women, purportedly Taylor Swift fans, wearing “Swifties for Trump” T-shirts. Among them was a manipulated “Uncle Sam” poster featuring Swift’s face, urging viewers: “Taylor wants you to vote for Donald Trump.” Trump’s caption? “I accept!”
While a couple of the images did feature a real Trump supporter, it was quickly revealed that the vast majority, at least fifteen, were fabricated using artificial intelligence. These digitally created personas were initially posted on X (formerly Twitter) by prominent pro-Trump accounts, some of which have even shared guides on utilizing generative AI tools on their blogs. These accounts later acknowledged that the images reposted by Trump were indeed AI-generated, labeling them as satire in some instances.
This orchestrated social media activity is part of a broader campaign by pro-Trump factions to falsely suggest a growing wave of support for Trump amongst Taylor Swift’s massive fanbase, known as “Swifties.” This is despite Swift’s public endorsement of Joe Biden in the 2020 election and her sharp criticism of Donald Trump. The reality is that there’s scant evidence of any significant shift in Swiftie allegiance towards Trump. As one of the accounts responsible for the AI images conceded, “There is no Swifties For Trump movement — but there should be,” highlighting the aspirational and fabricated nature of the campaign.
This tactic isn’t isolated to Swift’s fanbase. Similar AI-generated content has surfaced, falsely portraying other Democrat-leaning demographics, such as Black voters, as Trump supporters. In contrast to the AI creations, some genuine efforts to showcase Swifties for Trump have emerged, albeit with minimal organic traction. For example, content featuring a Liberty University student, a real Trump supporter and Swift fan, gained little attention until amplified by prominent pro-Trump accounts and eventually Trump himself. Similarly, TikTok videos of women claiming to be Swift fans voting for Trump have been reshared by these accounts, attempting to manufacture a narrative of widespread Swiftie support for Trump. However, these videos garnered significantly less viewership on their original TikTok platform compared to their reposted versions by Trump and his influential supporters.
The Swift episode is just the latest in a series of instances where AI-generated images of the pop star have gone viral. Earlier this year, sexually explicit deepfakes of Swift circulated widely, violating platform guidelines. Other AI-generated images falsely depicted her as a Trump supporter around the Grammy Awards, further illustrating the misuse of AI to create misleading content involving high-profile figures.
Throughout his political career, Trump has a history of sharing inflammatory memes and content created by his online base. While initially using Twitter, he migrated to Truth Social after his ban from the former platform in 2021. Even after returning to X, Trump continues to utilize both platforms to disseminate such content, including a recent fake image of Vice President Kamala Harris in front of communist symbols.
The legal ramifications of these AI-generated images, particularly concerning publicity rights and the unauthorized use of likeness, are still unclear. Tennessee, Swift’s home state, recently enacted legislation to protect individuals from AI-driven unauthorized likeness exploitation. While some accounts attempt to frame these AI images as satire, the lack of clear disclaimers in all instances raises concerns about potential deception and the spread of misinformation.
David Greene, a civil liberties expert at the Electronic Frontier Foundation, notes the potential legal overlap with publicity laws, especially when false endorsement is implied. While traditionally applied to product endorsements, this case highlights the evolving landscape where political endorsements and public perception are also being manipulated through AI-generated content. Ultimately, as Greene succinctly puts it, this tactic, regardless of the technological sophistication, boils down to “good old low-tech lying” amplified by the reach of social media and the deceptive potential of AI.
