Washington
CNN
—
Taylor Swift didn’t endorse former President Donald Trump over the weekend, Ryan Reynolds wasn’t photographed wearing a pro-Kamala Harris shirt, and the Communist Party of America has never endorsed President Joe Biden’s now-disbanded campaign.
But these false claims about the 2024 campaign and similar false endorsement posts are proliferating on social media as the election approaches, said researchers at the News Literacy Project, a nonpartisan education group that launched a new database Thursday documenting more than 550 instances of election-related misinformation.
The latest and most flagrant example of these false claims came on Sunday, when Trump shared a post on his platform, Truth Social, that included an image created using artificial intelligence with a caption suggesting growing support from Swift fans who call themselves “Swifties for Trump.” In response to the pop icon’s implicit endorsement, Trump wrote, “I accept!”
Swift, who has previously slammed Trump for “stoking the fires of white supremacy and racism” during his presidency, supported Joe Biden in 2020 but is yet to endorse a candidate in the 2024 presidential election.
One image in the collage posted by Trump showing Swift dressed as “Uncle Sam” was clearly doctored, but another image showing a young woman at the rally appeared to be real.
Other images purporting to show crowds of smiling fans congratulating Trump contain some of the hallmarks of AI-generated images, said Lucas Hansen, co-founder of CivAI, a nonprofit that raises awareness about the capabilities and dangers of AI.
Hansen said the images are “super airbrushed and high quality” and “everyone looks really cool,” and that they utilize the “strong bokeh and background blur” that is common in AI-generated images.
The News Literacy Project said it launched the misinformation dashboard on Thursday because it believes widespread misinformation poses an “existential threat to democracy” and is best examined by bulk analyzing hundreds of examples rather than through individual fact-checking.
The regularly updated database tracks several categories of political disinformation, including conspiracy theories, lies about candidates’ policy positions and fake endorsements, but the group does not measure how many times these viral posts are shared.
According to data provided exclusively to CNN, roughly 1 in 10 viral posts analyzed by the News Literacy Project contained fake endorsements or public snubs from celebrities, including NFL quarterback Aaron Rodgers, actor Morgan Freeman, musician Bruce Springsteen and politicians such as former first lady Michelle Obama.
According to the database, posts mentioning these four numbers have been viewed at least 10 million times.
The researchers found separate posts circulating simultaneously on social media purporting to accuse the same public figures of endorsing candidates, highlighting the chaotic and misleading environment users are encountering.
“As a general rule, if you see a public figure wearing a T-shirt with an obviously political message, there’s a good chance it’s fake,” said Dan Ebon, senior manager at the News Literacy Project.
Fake recommendations are proliferating on social media as tech platforms dismantle guardrails and moderation policies designed to curb the spread of dangerous misinformation.
The changes have been most visible at X, formerly known as Twitter, since billionaire Elon Musk bought the company, dismantled an internal team that worked to stop the spread of election misinformation and reinstated banned accounts of prominent conspiracy theorists and extremists.
Experts say the problem is exacerbated by X’s AI-powered chatbot, Grok, which has already drawn the ire of election officials for spreading false information about Harris’ qualifications for the 2024 election. Last week, X allowed users to use Grok to create AI-generated images from text prompts, sparking a flood of fake content about Trump and Harris.
“Going forward, Grok is likely to become one of the primary sources of this type of imagery because it produces high-quality imagery, is easily available, and is purposely built to have a low rejection rate,” Hansen said, adding that he was able to use Grok to create a “Swifties for Trump” image that looks very similar to the one Trump shared.
Company X did not respond to a request for comment about creating misleading images about political candidates.
Meta, which owns Facebook and Instagram, has also cut some staff from its election integrity team, CNN previously reported. The social media giant said last year that “no other technology company is working harder or investing more in protecting our elections,” and announced it would require political advertisers to disclose if they use AI to create or alter images.
But even as some online platforms work to label AI-generated images and fact-checkers debunk the latest viral lies, the proliferation of fabricated content could have consequences.
“When you see these false stories repeatedly that exaggerate the popularity of a candidate, it can stick in your mind as you scroll through your feed, even if you know it’s not true,” Peter Adams, senior vice president of research at the News Literacy Project, told CNN.
While the release of publicly available AI tools has made it easier to create misleading claims, including many of the false advertisements identified by the News Literacy Project, other fake images circulating on social media used rudimentary Photoshopping.
Adams said the use of AI to manufacture political disinformation “has not been as widespread and pernicious as people initially feared”, adding that traditional methods of doctoring images and videos are currently “a lot cheaper but just as effective”.
