Artificial intelligence–generated and digitally manipulated images are being used on social media to falsely associate prominent US politicians with the late convicted sex offender Jeffrey Epstein, according to disinformation researchers.
The disinformation watchdog NewsGuard said on Tuesday that at least seven fabricated images have circulated online, collectively drawing more than 21 million views on the Elon Musk-owned platform X, highlighting how AI-driven falsehoods are increasingly blurring the line between fact and fiction.
The images emerged amid renewed public scrutiny following the US Justice Department’s release last week of more than three million documents, photos, and videos linked to its investigation into Epstein, who died by suicide in custody in 2019.
While the Epstein case has involved high-profile figures—including Britain’s former Prince Andrew, American intellectual Noam Chomsky, and Norway’s Crown Princess Mette-Marit—others now being targeted online have no connection to the scandal.
Among those falsely implicated are New York Mayor Zohran Mamdani and former Republican presidential candidate Nikki Haley.
According to NewsGuard, conservative social media users circulated three images that purported to show Epstein posing with Mamdani, who appears as a child in the fabricated visuals. Two of the images also depict Mira Nair, the mayor’s mother and an internationally acclaimed filmmaker.
“These images are AI-generated fakes,” NewsGuard said.
The watchdog reported that analysis using Google’s AI tool Gemini detected a SynthID, an invisible watermark designed to identify AI-generated content.
One of the images gained widespread attention after conspiracy theorist Alex Jones shared it on X in a post that drew more than 1.5 million views, claiming that Musk’s AI chatbot Grok had verified it as real. Disinformation experts have long cautioned against relying on AI chatbots as fact-checking tools, citing frequent inaccuracies.
Fabricated emails and altered photos
Separately, a screenshot of an email purportedly sent by Nikki Haley to Epstein circulated online. The message claimed Haley had requested a flight from Epstein while traveling with children.
However, searches of the Justice Department’s Epstein files found no record of such an email. NewsGuard also identified inconsistencies in the screenshot, including an incorrect date: January 7, 2014, which was a Tuesday, not a Saturday as shown in the image.
Haley did not immediately respond to AFP’s request for comment. In a post on X last July, she had publicly urged President Donald Trump’s administration to release the Epstein files.
In another case, social media users in Latin America shared an image allegedly showing Epstein seated beside Venezuelan opposition leader María Corina Machado at the Hampton Classic Horse Show in 2002. Using reverse-image searches, NewsGuard determined the image was a digitally altered version of a real photo showing Epstein with a wealthy American businessman.
A recurring pattern
Researchers say such online fabrications are becoming increasingly common. Last year, similar AI-generated images attempted to falsely link Mark Carney, then a candidate for the leadership of Canada’s Liberal Party, to Epstein and his associate Ghislaine Maxwell. AFP fact-checkers later found strong evidence that those images were also AI-generated.
Disinformation experts warn that as generative AI tools become cheaper and more accessible, false associations involving public figures are likely to increase—particularly around emotionally charged scandals—making verification more difficult and misinformation harder to contain.
