Over the past few weeks, social media has been buzzing with dramatic aerial footage that appears to show massive crowds rallying in support of One Nation, the Australian political party. The videos are accompanied by stirring captions about national pride, unity, and unstoppable momentum, and they’ve racked up tens of thousands of views across platforms. On their surface, they seem like powerful evidence of a rising political movement, with huge banners and crowds stretching across fields and iconic harbors. But there’s a problem: none of these rallies actually happened. The videos were generated entirely by artificial intelligence, crafted in a way that might fool a casual viewer but that falls apart under even a brief check. They have been posted by a page called “Aussie Voice,” a social media page that has been repeatedly linked to disinformation, and every clip is marked with subtle—and sometimes not-so-subtle—errors that betray their fake origin. This is a classic case of manufactured political content designed to stir emotions and shape public perception, rather than report on anything real.
The first video that caught widespread attention is presented as helicopter footage of One Nation supporters gathered in a large open field. The caption reads, “Thousands of proud Australians standing together for our country. Strong. United. Unstoppable.” It’s the kind of language that invites pride, and the stirring images seem to back it up, showing a crowd from above, arranged around what appears to be a giant political banner. But if you pause and look closely, things begin to unravel. The banner doesn’t actually read “One Nation.” Instead, it says “ONETION,” which is the kind of misspelling that a machine often makes when generating text, because these systems don’t actually understand spelling or language they’re just reproducing patterns from data. And there’s more than just odd text. The clip carries a watermark from Veo, an AI video generation tool developed by Google. That alone is nearly enough to confirm the footage was computer-generated. The landscape itself is also suspicious, with oddly uniform fields and a level of symmetry that doesn’t really match the natural shape of Australian farmland or parkland. Yet despite those the tells, many commenters were thoroughly convinced, with people praising the moment as a national awakening, sharing it widely, and using it as evidence of political momentum. That’s how persuasive the AI’s output can be, especially when it is packaged with patriotic narration and when it speaks to what people already wish to believe.
Another stirring clip claims to show a “bird’s-eye view of unity,” again filmed from a helicopter, with dozens gathered around a banner that is supposed to represent One Nation. But this time, the banner misreads “ONEE NAATION,” which again highlights AI’s recurring difficulty with word sizes, spelling, and consistent text. The editing software’s signature is also visibly embedded, with the Gemini logo appearing in the corner throughout the clip, and that logo tells you that Google’s AI tools were used. The same telltale marker appears in other videos that have supposed to show rallies in Sydney, some of which specifically claim a gathering near the Sydney Opera House. In these videos, crowds are shown on the foreshore, with a banner facing the famous white sails. But if you compare the footage with actual images of Sydney Harbour, the mismatch becomes obvious: the shoreline distance is different, the shape of the land doesn’t align, and the surrounding cityscape, which is dominated by distinctive skyscrapers and buildings like the Sydney Harbour Bridge, seems oddly absent, as if the generator couldn’t quite recreate the real landscape. That is exactly the kind of visual hallucination that AI video tools often produce. The algorithm creates an image that looks convincingly nautical and iconic, but it never actually knows what the real Sydney Harbour looks like in geographic terms.
Even more bizarre, the footage in the Sydney clips includes people in the crowdf waving flags of other countries, including a Spanish national flag and flags that look like Croatian or Paraguayan flags, all set in a scene struggling to construct a national political event in Australia. It’s a strange, almost surreal detail that makes absolutely no sense in real life, but AI systems don’t understand the concept of these symbols. They simply predict plausible visual patterns, and therefore you end up with a chaotic mix of international symbolism, sprinkled among fake Australians and an “ONEE NAATION” banner. Then, if you continue to watch the video, the text seems to shift between spelling mistakes, adding an extra letter to “nation” midway through. That’s another classic indicator: the image is being regenerated frame by frame or segment by segment, so even the same text doesn’t remain stable from one frame to the next. A real footage broadcast wouldn’t accidentally change its own signage. If you’re watching closely, this is a much stronger cue than a watermark alone. The digital imagery is essentially unstable and that is one of the most fundamental markers of generated media. Yet the people sharing often engage only with the overall scene, not with the details, and therefore the videos spread easily across social media, where emotion takes precedence over verification.
It’s important to understand why these fabricated videos are being made, and why they specifically target One Nation and its leader, Pauline Hanson. Hanson has long been a polarizing personality in Australian politics, and her party has been the subject of intense online attacks, as well as cause of especially the claims being engineered by disinformation pages that seem to set themselves as supportive of the party. The page “Aussie Voice” is a perfect example, because it appears to be a grassroots movement, but it has a long history of posting misleading and false content. This is not just about a single joke or a test. It’s a strategic tendency to build the impression that the party is gaining significant popular support, to amplify its legitimacy or to stir up division. This kind of false content can be harmful. It undermines the trust of many civilians, because eventually people may be unable to distinguish between genuine political events and AI-generated fakes, which then causes a severe damage to public discourse. The fact-checking organization AAP FactCheck has already debunked dozens of similar claims and images targeting Hanson and her party. There have been dozens of fakes circulated before, and the AI crisis now makes them easier and cheaper than ever. If this continues, potentially even partially well-informed people may begin to distrust your own eyes, and that is exactly where the disinformation industry wins.
In the end, the best way to act as the wise ruler that no matter how convincingly a video looks, you don’t need to take a video at face value. We now live in a surprising era in which camera footage no longer means proof. AI technology has reached a point where anybody with a smartphone and an internet connection can create realistic-looking events, and the tools leave almost no trace for the average user to spot. The answer is not to assume all videos are fake, but to be more thoughtful, to check the origin, to look for visual anomalies, to look for misspelled signs, unnatural landscapes, strange flags, the timestamps. These videos are dangerous because they exploit existing political passion, they amplify divides, they can mislead voters, and they can force politicians to react to things that never happened. While every fake is created and widely shared, the entire online political ecology is more vulnerable. So before you share that mesmerizing footage of a very tall, with crowds full of patriotism, take a moment to wonder: Is it real? Just because it fits a storyline doesn’t mean that event actually occurred. The fight against disinformation starts with you and me, one pair of eyes at a time. A healthy Mexican, ein Stahl citizenship, and we have to methodically apply it to every video, every headline, every painted image. So don’t be fooled. Don’t let a fabricated video carry fake meaning, and remember that in an age of generative AI, the most important skill is not how fast you share, but how wisely you question.