Donald Trump’s announcement in the early hours of Saturday morning that US troops had captured Venezuelan president Nicolás Maduro and his wife, Cilia Flores, was followed almost immediately by a wave of disinformation online.
Across TikTok, Instagram, and X, users circulated old clips and AI-generated material as if they showed the operation, the arrest, or reactions in Caracas. The result was a fast-moving example of how major global incidents can become confused by recycled footage and synthetic media before basic verification catches up.
What Was Announced
Trump wrote on Truth Social in the early hours of Saturday morning: "The United States of America has successfully carried out a large scale strike against Venezuela and its leader, President Nicolas Maduro, who has been, along with his wife, captured and flown out of the Country,"
Hours later, US attorney general Pam Bondi announced that Maduro and his wife had been indicted in the Southern District of New York. According to the source article, the charges were narco-terrorism conspiracy, cocaine importation conspiracy, possession of machine guns and destructive devices, and conspiracy to possess machine guns and destructive devices.
Bondi wrote on X: "They will soon face the full wrath of American justice on American soil in American courts,"
Those official statements became the starting point for a much larger online information scramble. Within minutes, posts began presenting images and videos as evidence of the capture, the assault, or public reaction, even when the material did not match what users claimed it showed.
AI Images Became Instant Evidence
One of the most widely shared examples was an image that appeared to show two US Drug Enforcement Administration agents standing beside Maduro. It spread across multiple platforms soon after the news broke.
WIRED analyzed the image using SynthID, a technology developed by Google DeepMind that claims to identify AI-generated images. The analysis indicated that the image was likely fake.
Google’s Gemini chatbot wrote after analyzing the image: "Based on my analysis, most or all of this image was generated or edited using Google AI," It added: "I detected a SynthID watermark, which is an invisible digital signal embedded by Google's AI tools during the creation or editing process. This technology is designed to remain detectable even when images are modified, such as through cropping or compression."
The fake image was first reported by fact-checker David Puente. The source article also says that X’s AI chatbot Grok confirmed the image was fake when several X users asked about it. But Grok also falsely claimed the image was an altered version of the arrest of Mexican drug boss Dámaso López Núñez in 2017.
That detail matters because it shows a verification problem with two layers. The first layer is the fake image itself. The second is the possibility that an AI assistant may correctly identify something as false while still offering a wrong explanation for where it came from.
Old Footage Was Recast as New
The misleading posts were not limited to AI-generated images. Some users shared old videos and described them as current footage from Caracas on Saturday.
The source article describes this as a routine pattern after major global incidents. It points to the outbreak of the Israel-Hamas war in October 2023 and the US bombing of Iranian nuclear sites last summer as other moments when disinformation spreaders shared old footage as if it showed breaking events.
In this case, pro-Trump influencer Laura Loomer was among dozens of people who shared footage showing a poster of Maduro being taken down. She wrote on X: "Following the capture of Maduro by US Special Forces earlier this morning, the people of Venezuela are ripping down posters of Maduro and taking to the streets to celebrate his arrest by the Trump administration."
According to the source article, the footage was originally taken in 2024. Loomer eventually removed the post.
Another example came from an account called "Defense Intelligence," which posted a video shortly after Trump announced Maduro’s capture and claimed it showed footage of the US assault on Caracas. The post had been viewed on X over 2 million times. The footage had originally been posted on TikTok in November 2025, and at the time of publication the post remained on X.
Platform Moderation Is Part of the Story
The source article connects the spread of these posts to a broader shift in how social platforms handle major events. In recent years, major global incidents have triggered large amounts of disinformation as tech companies have pulled back efforts to moderate their platforms.
That weaker moderation environment creates an opening for accounts looking to gain engagement and followers. When an event is dramatic, fast-moving, and politically charged, misleading content can travel quickly because it appears to give viewers what they want most: a visual answer before verified information is available.
The material described in the source article followed several patterns:
- Old footage was presented as if it had been recorded in Caracas on Saturday.
- AI-generated images were framed as documentation of Maduro’s arrest.
- AI-generated videos were made from images that purported to show the arrest.
- Posts moved across TikTok, Instagram, and X before platforms publicly responded.
Some TikTok examples of apparently AI-generated videos reached hundreds of thousands of views within hours of Maduro’s capture. A number of those TikTok videos appeared to be based on AI-generated images originally posted on Instagram by a digital creator named Ruben Dario and viewed over 12,000 times. Similar videos also appeared on X.
X, Meta, and TikTok did not respond to requests for comment.
The Verification Gap
The Maduro posts show how quickly a news event can become a test of verification systems. A single announcement produced a mix of official statements, synthetic visuals, old footage, AI chatbot responses, and viral claims that did not all point in the same direction.
WIRED earlier reported that ChatGPT strongly denied that Maduro had even been captured when asked about the event on Saturday morning. Meanwhile, SynthID helped identify a likely fake image, and Grok both confirmed that the image was fake and offered a false claim about its origin.
For readers, the practical lesson is narrow but important. A viral image or video attached to a breaking event is not proof by itself. The source article shows that even when content looks specific, spreads widely, and appears soon after a major announcement, it may still be AI-generated, old, mislabeled, or incorrectly explained by the tools used to check it.
In the first hours after a major incident, speed often favors the most shareable version of events, not the most accurate one. That is the opening disinformation used after Trump’s announcement about Nicolás Maduro’s capture.