Try to imagine the clip if you haven’t seen it yet. It is a bright afternoon in Palm Beach County, Florida, and a small business jet has somehow made its way into the frame. There is nothing graceful about its descent. The nose of the aircraft drops, and the plane plummets straight toward a residential neighborhood, as if it has been thrown out of the sky. It slams into the pavement at full speed, cratering the asphalt in a burst of force. The metal crumples, and in the same instant the aircraft dissolves into fire and dark smoke. The whole sequence lasts only a few seconds, but those seconds are so violent, so cinematic, that they seem impossible. If you watched it online, you might have scrolled back, wondering whether it was a clip from a film or a custom-made special effect. That initial feeling of disbelief is exactly what social media users seized on. Within hours of the crash, posts began circulating that claimed the video was not a recording of real life at all. “The Video You’re All Seeing From The ‘Plane’ Crash Was 100% Ai Generated SOMEONE IS LYING TO US !!!!!! That Was Not An Airplane,” one X user insisted on Nov. 1. The post included what looked like proof: screen recordings of the video being run through an AI-detection tool, which appeared to flag the plane as synthetic. Other accounts repeated the claim, some with fewer exclamation marks but with the same message. The video was fake, they said. The crash was fake. The entire event was a lie. But as PolitiFact’s reporting made clear, the only lie was the one being told about the video. The crash happened. The video was real. And the story of how a tragedy became a conspiracy theory is a cautionary tale about trust in the age of AI.
Let’s start with what actually happened, because the facts matter. On Friday, Nov. 1, at about 12:51 p.m. Eastern Time, a Learjet 60 crashed into a neighborhood near Cresthaven, Florida. The plane had departed from Austin, Texas, and was headed to the Bahamas. It was a business jet, not a military drone, not a CGI object, not a staged stunt. U.S. Transportation Secretary Sean Duffy confirmed on the same day that everyone on board was killed. The National Transportation Safety Board immediately launched an investigation. At a press conference the next day, board member Tom Chapman said the aircraft had experienced turbulence and had encountered ice. Those early findings were not a final explanation, and Chapman was careful not to assign blame. Timothy Monville, the NTSB investigator leading the on-site effort, described a complex and grisly scene. “We have a lot of work to do on scene,” he said. “This will be a several-day process, and the biggest obstacle that we’re going to have is recovering the major wreckage from the impacted crater on the roadway.” That crater was not a visual effect. It was a hole in a Florida street, surrounded by shattered debris and flames. The video that had gone viral showed that same crash from the perspective of a camera mounted on a car. Its original source was a digital creator and West Palm Beach resident named Aki Kuusisto, who posted the clip to Facebook. In the comments, he expressed the kind of grim irony that only a witness to something surreal could muster: “This was not quite what I had in mind when I installed the dash cam.” Kuusisto’s footage was soon shared by The Associated Press, CNN, CBS News, and BBC, each outlet doing its own verification. The source was real. The location was real. The time was real. The wreckage was real. None of this left room for the claim that the plane was a hallucination generated by artificial intelligence.
So why did so many people believe the AI claim? Part of the answer lies in a tool called Hive Moderation. The viral X posts showed screen recordings of the dashcam video being uploaded to Hive, an AI-detection service commonly used to identify synthetic media. In those recordings, the tool reported that it had detected AI-generated content, and the concern seemed specifically centered on the plane. For a viewer scrolling quickly, that looked like conclusive evidence: an algorithm had examined the video, and the algorithm had said fake. But Hive is not infallible, and a single result from an automated classifier is not proof of anything. PolitiFact has used Hive in its own reporting, but always with an understanding of its limitations. These tools operate by recognizing statistical patterns, not by understanding what they see. They are trained on datasets of known AI-generated images and videos, and they can be thrown off by common internet mutations: screen-recording artifacts, compression artifacts, transcoding, reduced resolution, and the loss of metadata. The version of the video circulating on social media was exactly the kind of degraded copy that confuses classifiers. When PolitiFact tested the original footage directly from Kuusisto’s dashcam, Hive did not flag it as AI-generated or manipulated. That result is not a perfect rebuttal, but it matters: the same tool that supposedly exposed the hoax failed to do so when given the authentic file. The screenshots that fueled the conspiracy were misleading in another way as well. They showed only a moment of analysis, without context or explanation. They did not show how Hive’s confidence ratings can vary across frames, nor did they acknowledge that the tool has a false-positive problem, especially with videos that have been screen-recorded and re-uploaded. In the end, the “AI detector” was not a detective. It was a prop.
The willingness of so many people to accept that prop as proof is a symptom of a broader cultural condition. Darren Linvill, co-director of the Media Forensics Hub at Clemson University, spends his days studying disinformation. He told PolitiFact that skepticism is entirely appropriate in an age when generative AI can produce photorealistic fakes. But he said the reaction to the Palm Beach video illustrated something more troubling: a phenomenon known as the “liar’s dividend.” The term refers to the way people become so accustomed to being lied to, so conditioned to expect deception, that they no longer recognize truth when it is right in front of them. To put it plainly, the liar’s dividend is the profit that liars earn from the exhaustion of honest people. Linvill put it this way: “This is when people are so accustomed to being lied to, they don’t know the truth when it is presented to them. That is what is occurring with the video of the Palm Beach crash.” The phrase captures the strange paradox of the AI era. Deepfakes are dangerous because they can make people believe things that never happened. But the existence of deepfakes also makes it possible to dismiss things that really did happen. Every real crime scene, every real disaster, every real interview can be smeared with the accusation that it was manufactured. The public, burned too many times by hoaxes and exaggerations, starts to doubt everything. In doing so, it grants a kind of immunity to the actual deceivers, because no standard of evidence can satisfy someone who does not believe evidence is possible. That is what the Palm Beach crash became in the hands of the viral AI claim: not a tragedy, but a Rorschach test for the internet’s broken relationship with reality. The people who died on that plane were real, with names and families and histories. But in the misinformation ecosystem, they were reduced to pixels that allegedly looked “off.” That reduction is more than an intellectual error; it is a moral failure, one that feeds on grief and turns a public tragedy into another partisan battlefield.
Why did a dashcam video look so unreal that viewers assumed it was CGI? The truth is that everyday life rarely offers such a dramatic image. Plane crashes unfold far from cameras, and when they are captured, they are often blurry and hard to make out. Here, the camera was stationary, the day was bright, and the plane was moving fast enough to create a terrifying visual arc. It looked like a movie because movies have trained us to see disaster through cinematic language. But the reason it was captured at all is more mundane: people have recording devices everywhere now. Dash cams hang from windshields. Doorbell cameras watch quiet streets. Smartphones sit in pockets, always ready. It would have been more surprising if a crash in a residential neighborhood had not been captured from some angle. The abundance of recording technology, however, also creates new opportunities for abuse. PolitiFact’s reporting noted that one of the X accounts that helped spread the AI claim had been amplified by bot accounts linked to Russian Storm-1516, a disinformation campaign that uses Russian and non-Russian influencers to promote false claims in the United States and abroad. That is not to say that every person who doubted the video was a foreign agent or a bot. Many were simply misled, confused, or anxious about the future of media. But the amplification pattern reveals something important: accusations of AI-generated content are not always organic. They can be seeded, boosted, and exploited by actors who want to destabilize public trust. When nothing can be believed, everyone is free to believe whatever suits them, and that is a dangerous environment for any society. Linvill’s advice is simple and durable: get news from reliable sources, think before sharing, and do not let a single tool, let alone a screenshot of a tool, override the demonstrated facts. Verification is not glamorous, but it is the only antidote to the liar’s dividend.
In the end, PolitiFact’s ruling was characteristically blunt: Pants on Fire. The claim that the video was “100% AI generated” was not just inaccurate; it was a reckless attack on the truth. The video came from a real person. It was verified by multiple independent news organizations. It matched the details of a real NTSB investigation. The source file, when tested by PolitiFact, did not produce the same AI-generated flag that the viral screenshots did. The expert who explained the viral claim was not making an excuse; he was issuing a warning about the consequences of mass distrust. None of that means every video should be believed instantly. Deepfakes are improving, and the threat of synthetic media is real. But the existence of fakes does not make everything fake, and the proper response to a video like this one is a clear-eyed process of verification, not a reflexive crawl toward conspiracy. The people who died in the Palm Beach crash deserve to be remembered as victims of a real tragedy, not as extras in an internet hoax. Their families deserve to mourn without having to watch strangers argue about whether the plane was made of pixels. And the public deserves an information environment in which a terrible event can be acknowledged, investigated, and mourned without first being subjected to a trial by screenshot. The liar’s dividend is real, but it is not inevitable. We can choose to believe verified sources. We can choose to be humble about what we don’t know. We can choose to say, when a video of a plane crashing looks like a special effect, “That horror was real, and it matters.” The dashcam was not lying. The crash was not CGI. The only fabrication was the claim that the truth was fake.