If you happened to check social media in the aftermath of the Trump administration’s latest drop of Epstein files, you may have encountered the following claim: New York’s City’s recently elected socialist mayor, Zohran Mamdani, is actually the secret biological son of convicted sex trafficker Jeffrey Epstein.
Supposed photos depicting Mamdani at various ages, his mother (filmmaker Mira Nair), and Epstein were shared tens of thousands of times, including by prominent right-wing influencers. Sandy Hook conspiracy theorist Alex Jones even posted on X that there was a “major investigation” underway to determine the mayor’s parentage.
It was all bullshit — a dumb, AI-fueled conspiracy theory apparently based on an email between a publicist and Epstein that mentioned Nair as an attendee at a promotional screening afterparty for her 2009 film Amelia, which was hosted at the home of Ghislane Maxwell. Mayor Mamdani was born in 1991.
The episode laid bare the newest, most rapidly evolving frontier in news misinformation: the collision between the changing manner in which global audiences consume information, and an artificial intelligence revolution that allows any individual to generate increasingly detailed and realistic content in just a few seconds.
Nowhere has this been more apparent than throughout the Epstein files saga. For the better part of a year, the Trump administration has attempted to delay, defer, and deny public access to materials related to the case. After being forced to release the files by an act of Congress, the material has been released in dumps of hundreds of thousands, even millions of documents at a time — including six million of them last week. The onslaught of emails, business logs, photos, testimony, videos, court documents, and personal correspondence has spread coverage of the files so thin that it’s near impossible to parse. Adding to the confusion is that those already muddy waters are being regularly stirred up by AI misinformation slop.
In March of last year, researchers debunked a photo purporting to depict Canadian Prime Minister Mark Carney, actor Tom Hanks, and Epstein’s convicted accomplice Ghislane Maxwell on the beach together as AI-generated. In January, a viral audio clip presented as a recording of Trump berating former Rep. Marjorie Taylor Greene over her support for the release of the files was actually generated by OpenAI’s video generating software Sora. After a tranche of Epstein documents containing mentions of both Trump and former President Bill Clinton became public, an AI-generated video of Trump patting and kissing Clinton’s crotch went viral on several social media platforms. A slew of celebrities, including Oprah Winfrey, former Vice President Kamala Harris, and Open AI CEO Sam Altman have been caught up in the AI slop vortex swirling around the release of the Epstein files.
In the case of Mamdani, major conservative figures circulated AI-generated images purporting to show him as an infant on a beach vacation, in his mother’s arms, with Epstein looming behind them. Other AI-generated images claimed to show a young Mamdani alongside his mother, Epstein, and other prominent figures named in the files. According to BBC Verify, the latter images seem to have originated from the parody account “@DumbFckFinder.” Despite containing an AI watermark, X’s AI chatbot Grok told at least one user that the photo was authentic.
“Somebody has to say it. There is a very real possibility Zohran Mamdani is Jeffrey Epstein’s biological son,” a pro-Trump account wrote in an X post with over 2 million engagements and 90,000 likes.
“Grok says this photo of the young future, New York City mayor Zohran Mamdani with Jeffrey Epstein, Bill Gates, Bill Clinton and others is real,” Alex Jones wrote in one of his posts. “I’m about to break huge news on this topic in the next few hours.” No news was broken, and fact-checking notes later added to his posts clarified that the images had been made ”with Google Nano Banana.”
It’s not particularly surprising given the extent to which the Trump administration and the MAGA movement have used generative AI to create the reality they wish existed. The impulse has been particularly visible in Minneapolis, where in just the past few weeks the administration has — through its official social media accounts — distributed a manipulated image of lawyer and activist Nekima Levy Armstrong during her arrest, falsely showing her in tears. The Department of Homeland Security, ICE, and Border Patrol regularly share AI-generated content depicting immigration agents as jingoistic — and decidedly Nordic-looking — warriors for the motherland. The pattern is repeated across the administration, and Trump’s near-daily barrage of Truth Social posts and reposts are typically clogged with AI slop depicting him as a king, superhero, or literally dumping shit on his rivals.
The integration of AI into the public’s interfacing with news events is a multifront misinformation war on people who just want the news. From Google AI summaries serving up ludicrously incorrect answers to queries, to news outlets publishing fake books and authors after using AI to create a reading list, to the ease with which AI can create newscasts out of thin air or manipulate the image of a news personality into saying something completely fake.
There is no need to invent personalities, or connections, or photos related to the Epstein files; the case itself is horrifying enough. The testimonials given by survivors are harrowing and contain plenty of leads worthy of investigation, as do the now public communications between Epstein and his wide network of contacts and associates. The Epstein saga as a whole calls for a sober, independent, detailed investigation into the multiple failures of the American criminal justice system that allowed him and his accomplices to continue abusing young women years after he was publicly known as a predator. No AI hallucination can even begin to imitate those horrors.
