The judge demanded that White House Press Secretary Karoline Leavitt remove her cross necklace, but soon he was humiliated. A man tried to yell at Vice President JD Vance in a cafe, but soon, he too was brought low. Barron Trump triumphed in a debate over a “liberal student” who bore a striking resemblance to the actress Emma Watson.
On YouTube, videos featuring these bizarre and overwrought tales are proliferating, claiming Trump-world figures bested their antagonists during micro-interactions: at restaurants, on the street, in little-known court cases, and on late-night talk shows. These videos often generate thousands of views and a flood of supportive comments—and they’re all completely fake.
YouTube hosts the fakes even as it backs legislation meant to fight the “rampant spread of unauthorized AI-generated deepfakes.”
The videos are part of an increasingly broad river of AI slop that runs through the internet, made to drive engagement—a torrent of low-quality, artificially-generated content that’s as clumsy as it is omnipresent. AI slop is especially prevalent on TikTok and Facebook, where it is often created by low-wage independent workers in developing countries using AI-generated content as a way to earn money while sitting near the bottom rung of a global economy.
In the wake of Trump’s second election, essentially-anonymous channels have churned out dozens of videos telling heartwarming stories about players in Trump’s world. The YouTube videos tend to focus on small interactions, presenting invented narratives where a MAGA celebrity—often press secretary Karoline Leavitt or Barron Trump, in a surprising number of cases—face some challenge and come out victorious. The videos are published by accounts with names like Elite Stories and Mr. Robe Stories, the latter of which says it offers “touching stories, stories about understanding, empathy and human rights equality.” Several of the channels have ads in their videos, meaning that the videos are being monetized.
Most of the channels add a disclaimer either in their channel description or in the videos themselves that the stories are fake. Mr. Robe Stories’ videos, for instance, offers a disclaimer in its videos’ first frames that said stories are fake. The disclaimer itself appears to AI-generated, with the clumsy and repetitive wording that’s a hallmark of machine-generated content. It reads, in full, “All stories shared on this channel are purely fictional and created for entertainment purposes, but we want a planet where people have the right to equality before the law, a society that is increasingly fair, developed and civilized. Any similarities to real events, individuals or organizations are fictional or coincidental and unintentional. The stories are not intended to refer to any real-life events, people or entities, especially no political or religious references.”
I located at least ten accounts specifically producing political AI slop focused on the Trump administration and family. While many of the accounts boast tens or even hundreds of thousands of subscribers, it’s difficult to tell how many of those followers are real humans. (Dozens of dodgy sites offer tools to buy YouTube subscribers, even though the practice violates the company’s terms of service.)
But plenty of these videos seem to be getting interaction from real people who seem to think the videos tell real stories, judging by the large volume of supportive and credulous comments, like “PAM WILL DEFEAT THAT JUDGE, GOD ALWAYS WINS!” And the videos can rack up a huge numbers of views in a very short period of time: the Elite Stories channel was created in October, and produced its first video in late November, but has already garnered more than 31 million views, according to the account’s info page. (Mother Jones contacted Elite Stories and several other accounts for comment, but did not receive any reply.)
Many of the accounts produce multiple videos a day. Similar content is replicated across the accounts—which could indicate the accounts are operated by the same person or people, or simply that what does well on one account is quickly copied. Whatever the explanation,highly specific themes are repeated often. Stories about cross necklaces, for instance, are especially popular: a video about a fictitious judge demanding that Attorney General Pam Bondi remove a cross necklace has generated 2.8 million views so far. (A substantially similar story on another account did less well, with only 109,000 views.) One about Karoline Leavitt being fined by a judge for wearing her cross necklace, before realizing she’s a “legal genius,” has been viewed some 89,000 times.
Leavitt often wears an eye-catching cross necklace, and anyone searching for news about her jewelry will more than likely stumble across an AI-generated video. (Another common, albeit slightly less successful theme, is Barron Trump seeking to adopt a dog, sometimes accompanied by an AI-generated blonde girlfriend.)
YouTube does have a misinformation policy, which bans “misleading or deceptive content that poses a serious risk of egregious harm.” Under that metric, there’s no “egregious” harm in an unknowable number of people thinking that, say, Barron Trump got a golden retriever.
The content also continues to exist on the site at the same time YouTube is supporting the No Fakes Act, a bipartisan piece of legislation working its way through Congress, which is meant to, as one of the bill’s authors, Rep. Madeleine Dean (D-Pa.) put it in a press release, address the “rampant spread of unauthorized AI-generated deepfakes without stymieing innovation or censoring free speech.” In their own press release backing the bill, YouTube’s vice president for public policy said the company’s support of the legislation “reflects our commitment to shaping a future where AI is used responsibly.”
In response to a request for comment, a YouTube spokesperson said it had deleted one of the pages, Elite Stories, “for violating our spam policy,” which “applies to all channels, regardless of whether their content is generated using AI tools.” The spokesperson added, “When a channel is terminated, it is against our Terms of Service to use another channel to circumvent that decision.”
Many of the video stories use an AI-generated narrator who speaks in a slow, professional cadence—like an actor reading an audiobook. They’re clearly meant to by turns enrage and soothe audiences, painting a picture ofinjustices that are, by the end of the video, made right. Some commenters have signaled they understand the videos are fake, but enjoy them anyway. “I have been listening to various stories concerning Keanu Reeves. So I decided to look into it and I found that they were just stories,” wrote one person on a Bondi cross video. “So now, when I listen to these things, the first thing I tell myself is, this is just for entertainment.”
From Mother Jones via this RSS feed