Shortly before midnight on Thursday, amidst the government shutdown, President Donald Trump shared a video on Truth Social. In the video, he is seen in the Oval Office seemingly tossing one of his red hats onto someone’s head while the song “Y.M.C.A.” by the Village People plays. Trump gestures and laughs.

The moment was fabricated: It was an evidently altered video designed to mimic a recent meeting between Trump and House Minority Leader Hakeem Jeffries. A few days prior, Trump shared a modified video of Jeffries and Senate Minority Leader Chuck Schumer, which also contained fake audio and presented Jeffries wearing an animated sombrero.

Democrats lambasted Trump. His allies laughed.

Regardless of opinions on the matter, Trump’s adoption of synthetic media has successfully captured attention, especially as meme-driven politics have surged on social media. This trend is apparent with many from the right and some from the left frequently engaging in internet-savvy humor. While AI videos are starting to enter this landscape, few politicians match Trump’s frequency in posting them.

Trump’s use of generative AI media seems to be accelerating. According to an NBC News analysis, he has shared numerous pieces of synthetic media, including AI-generated images and deepfake videos, on his Truth Social account since reoccupying the White House in January. Notably, about half of these posts were made in August and September, with many originating from others who initially shared them online.

Valerie Wirtschafter, a fellow at the nonpartisan Brookings Institution who examines AI in politics, remarks, “They go viral. They get attention. They’re this meme-ification of politics. They’re a little odd. But they get shared. They get eyeballs.”

Increasingly, the president has relied on fake images and videos — some likely AI-generated — to ridicule and challenge political opponents while also enhancing his own image. He has also used modified and AI-created media to comment on major national issues, such as revisions to the Cracker Barrel logo, and to hint at future government policies on topics like immigration and health care.

The AI content on Trump’s account falls on a spectrum from obviously fake mythmaking images, such as Trump standing next to a roaring lion, to content with a higher potential to mislead people, such as the fake video of Trump throwing a hat at Jeffries.

With the government shutdown expected to drag into next week, Trump posted four videos in four days that appeared to have been created or altered with technology, including one set to an alternate version of Blue Oyster Cult’s “(Don’t Fear) the Reaper,” with Russ Vought, director of the Office of Management and Budget, as the Grim Reaper, an apparent reference to plans for the mass firing of government workers.

The posts have an enormous online audience. They go out to Trump’s 10.8 million followers on Truth Social. And many of the posts are later shared to Trump’s other social media accounts, including on X, where he has 109.5 million followers. Some of the AI posts get tens of millions of views, according to X’s data, creating an incentive to post more of them.

The use of generative AI by Trump and his staff has spiked alongside the tech’s broader rise on the internet. Millions of people are downloading AI engines such as Google Gemini and OpenAI’s ChatGPT, creating waves of synthetic content for social media. Trump has presented himself as an ally of the AI industry, inviting tech CEOs to the White House and releasing an “AI Action Plan” to encourage development. And AI-forward company stocks are booming, pushing the stock market to all-time highs this week.

And it comes as generative AI continues to develop rapidly, with this week’s release of OpenAI’s Sora 2 putting advanced video generation onto millions of smartphones.

The White House said it stands by Trump’s sharing of AI-generated content.

“President Trump is the most memetic communicator in presidential history, and his adroit use of social media has been key to driving his Make America Great Again movement to the forefront of American politics and society,” White House spokesman Kush Desai said in a statement to NBC News.

The Trump administration has fully embraced the technology as part of its second-term communications strategy. Federal agencies frequently post memes and AI images on social media to play to Trump’s base, such as a cartoon depiction in March that appeared to mock a crying migrant who had been arrested.

“He’s incredibly transparent, as you all know,” White House press secretary Karoline Leavitt said in a briefing Wednesday.

“You hear from him directly on social media. He likes to share memes. He likes to share videos,” she said. “He likes to repost things that he sees other people post on social media, as well, and I think it’s quite refreshing that we have a president who is so open and honest, directly himself. Many a times, on Truth [Social], you are hearing directly from the president of the United States.”

Experts in AI said Trump is creating an alternate world for his followers.

Samuel Woolley, an associate professor of communication at the University of Pittsburgh, said that generative AI images and video can “bend reality” for people who consume them.

“They serve the goals of individuals and entities that are hoping to create a narrative that is alternative to scientific facts, or to create a brand-new version of what people should pay attention to and what is noteworthy,” he said.

While some of the images and videos, such as one in February about Trump ruling over Gaza, have drawn wide attention, others have received little notice in a frenetic information environment. Trump also posts plenty of unaltered images and videos alongside written messages, sometimes in his signature all-caps style.

In May, Trump shrugged off questions about an AI-generated image of him as pope. The image had appeared on his Truth Social account and on a White House X account, and Trump said he had nothing to do with it.

“They can’t take a joke?” Trump said at the time, referring to Catholics who said they were offended by the image. “Give me a break. It was just, somebody did it in fun. It’s fine. Have to have a little fun, don’t you?”

The White House has also dismissed the idea that internet jokes are somehow beneath the office, posting on X in July: “Nowhere in the Constitution does it say we can’t post banger memes.”

Of the dozens of generative AI posts examined by NBC News, about a third were videos and the rest were still images. Another dozen posts had images or video that appeared to use not artificial intelligence but older, cruder forms of digital manipulation, such as warping a photo of comedian Rosie O’Donnell to make her appear heavier.

As with the Trump-as-pope image, it’s not always clear who created the media and what software they used, although some of the images and videos include credits. Of the dozens of pieces of media, about two-thirds were ones posted directly to Trump’s account, while the rest were reposts of content from another account.

The tally does not include posts by other accounts such as the White House’s X account, which this year has posted AI-created portraits of Trump as a king, as Superman and as a character from “Star Wars.”

Some contain outright falsehoods. Last weekend, Trump posted a deepfake of himself and his daughter-in-law Lara Trump promoting a conspiracy theory about “medbeds,” a fake health cure-all popular in some online communities. The video was later removed.

Leavitt defended the video, saying, “I think the president saw the video and posted it and then took it down. And he has the right to do that. It’s his social media.”

Trump often employs AI to attack political rivals. A deepfake of Rep. Nancy Pelosi, D-Calif., shared by Trump on Sept. 8 uses a falsified voice to depict Pelosi admitting to a crime. A post on Sept. 3 presented Democratic Gov. JB Pritzker of Illinois and former New Jersey Gov. Chris Christie, a Republican, as sumo wrestlers. A post in July depicted Sen. am Schiff, D-Calif., in handcuffs, while Democratic Gov. Gavin Newsom of California is the subject of three posts.

Newsom, who also has recently embraced posting memes, struck back on Tuesday, with his press office posting an AI-generated video on X attacking Vance.

A Trump post on Truth Social from July 20 used AI to falsely show FBI agents arresting former President Barack Obama, sparking outcry. That video appeared near the height of a MAGA backlash over the Jeffrey Epstein files and drew condemnation from an Obama spokesperson, who called Trump’s claims that Obama rigged the 2016 and 2020 elections “bizarre” and “ridiculous.”

For Trump, the embrace of new media is an old skill, echoing his sway over New York tabloid newspapers in the 1980s through his use of Twitter, now X, when it was still new.

“He’s a master media manipulator, going way back,” said Michelle Amazeen, an associate professor of mass communication at Boston University who researches misinformation. “What he’s doing is dominating on social media, on television, in other news outlets, and it gives the illusion that this is the way things are now.”

Amazeen said AI-generated media may not be as popular as it appears. In a survey this year conducted by polling firm Ipsos for her research center at BU, 84% of respondents said such content should be clearly labeled and 81% said social media apps should be required to remove unauthorized deepfakes. (Trump’s AI posts are rarely labeled as such.)

“There’s a whole swath of people out there who are so turned off by this that they’ve tuned out,” she said.

Several of Trump’s AI posts relate to serious policy topics. On a Saturday in August, Trump posted a fictitious image of himself at the New York Stock Exchange’s famous bell next to a sign for something called the Great American Mortgage Corporation. That company does not appear to exist, but the post appeared to reference a proposal to sell the government’s stakes in mortgage giants Fannie Mae and Freddie Mac under a new entity with the ticker symbol MAGA. A sale hasn’t happened but remains a topic of discussion.

On a Saturday in September, Trump posted a cartoon image depicting him firing Federal Reserve Chairman Jerome Powell, also prompting news coverage.

Sometimes Trump’s intent just isn’t clear, said Daniel Schiff, an assistant professor of technology policy at Purdue University.

“Is this something that’s going to happen? Or something that they just might want to happen?” he said. “It can be a gray area.”

AI-created animals sometimes pop up on Trump’s feed, including at least three lions and one dove. And at least four AI-created posts include a reference to the long-running QAnon conspiracy theory.

Trump, though, is the main character on his Truth Social feed. When generative AI or other synthetic media appears on his feed, it has often included a fictitious depiction of Trump in some form, including: Trump as a pole vaulter, Trump as SWAT team member, Trump as a train conductor, Trump as an orchestra conductor, Trump as a character from the film “Apocalypse Now” and Trump as a member of the rock band Journey.

Schiff, who helps to run a database of political deepfakes and who is not related to the California senator, said it’s notable to see generative AI used to create so much “positive” content, after years of focus by researchers and journalists alike on deepfake scams or deepfake political attacks. He also said that Trump is the most widely popular subject for AI fakes across the internet, among fans and critics.

“The plurality of deepfakes are about Trump,” he said, excluding sexual deepfakes.