By Tatum Hunter
The internet has always had a problem with the truth. False rumors and smear campaigns have spread across social media for two decades, destroying reputations in their wake. But events over the last week suggest that what's coming is even worse: AI-generated video that supercharges online harassment.
Just days after an ex- JPMorgan employee filed a sexual assault and discrimination lawsuit alleging a female co-worker at the bank demanded to have sex with him and threatened his career advancement if he didn't, "deepfake" videos of the two parties spread across X and Instagram, racking up millions of views.
One clip shows the female executive dancing suggestively in an office while male co-workers shower her with dollar bills. In another, she caresses the ex-employee while he works at a desk. ("50 Shades of JP Morgan," the post says.) In a third, they are at the Met Gala.
A particularly alarming example depicts the two parties through the window of a cafe, smiling together over a glass of wine. The clip is insidious in part because it is not outrageous: It looks as if it captures a private moment. At the end of the video, the couple notice they are being watched and look worried, as if they have been caught.
JPMorgan said it conducted an internal investigation and found no support for the ex-employee's claims. The female employee's legal team said the accusations are false and that she and the ex-employee never had a romantic or sexual relationship. Both sides are gearing up for a possible trial. And yet, the images sweeping the internet are creating lurid fictional narratives about the two people at the center of the case, in some cases grabbing details straight from the plaintiff's complaint to paint a dubious portrait of everyone involved.
"We believe images," said Danielle Citron, a law professor at University of Virginia who studies online abuse and privacy. "They viscerally hit us."
This can be true even when the AI-generated imagery is glitchy, cartoonish or jokey. A 2025 study by Microsoft found that humans correctly guess whether an image is real or generated only 62% of the time -- slightly better than a coin flip. Because people struggle to spot AI images, it could influence our perception of real-world events, another 2023 study suggests.
The digital world is "taking the likenesses of these two people and using them in a way that has a life of its own," said Jeremy Carrasco, director at independent media watchdog Riddance, which helps people spot AI content.
Making AI images is easy. It only takes a minute or two to generate a fake video of a real person using a chatbot like X's Grok. As a result, this kind of media will likely play a growing role in the manufacturing of online scandals, said Renée DiResta, a researcher who studies internet shame campaigns.
One 2020 study found that including an image with a tweet more than doubled its reshares. When those visuals stoke outrage or invoke a deep-seated bias, onlookers are even more likely to engage. Content creators, who make money from views, are likely to add their own commentary or conjecture to any controversy that promises to bring in more eyeballs.
Citron said the unfolding incident at JP Morgan immediately made her think back to a defamation case the actor Johnny Depp brought against his ex-wife Amber Heard after she wrote a Washington Post op-ed that described herself as a "public figure representing domestic abuse."
That trial became the center of a monthslong online media circus. Major YouTube and TikTok creators -- most without any legal background whatsoever -- pumped out hundreds of videos dramatizing each step of the trial. The videos drew audiences in the millions, which impacted the public's perception of the case.
And that was before the era of AI videos, Citron noted.
There is very little established recourse for civilians who find themselves deepfaked. But a class-action lawsuit filed in January is taking aim at Elon Musk's xAI after the company's chatbot Grok allowed people to flood social media with millions of digitally altered nude images of real women and children. The lawsuit cites multiple grievances, including privacy violations, negligence and defamation. Ashley St. Clair, a content creator and the mother of one of Musk's children, also sued xAI this year over sexualized deepfakes depicting her.
The consequences of these lawsuits could be wide-ranging, according to Citron, potentially changing how tech companies view their own legal liability or, conversely, how safe victims feel speaking out after they're targeted with an AI smear campaign.
But the outcome won't change the incentives that make AI video useful for professional and amateur internet-posters -- in addition to those who seek to shape public opinion in much more powerful ways. "We're going to see more weaponization of this kind of synthetic lie," Citron said.
(END) Dow Jones Newswires
May 15, 2026 20:00 ET (00:00 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments