Published on 2026-07-25
What if the dramatic news photo shows something that never happened?
A fake Pentagon explosion photo briefly moved the stock market in 2023. The image looked convincing. The event simply never took place.
On the morning of 23 May 2023, an image of a large explosion near the Pentagon started moving fast online, shared in part by accounts with verified checkmarks. It looked exactly like the kind of photo that accompanies a real crisis: smoke, an official-looking building, the framing of a breaking-news screenshot. Markets twitched. Then the Pentagon and local fire officials said, plainly, that nothing had happened at their building that morning. There was no explosion. There was only a picture of one.
The picture was the whole event
That's what makes this category of fake different from an old photo reused with a new caption. There, at least, something real happened once, somewhere. Here, the image is the entire alleged event. No fire crews arrived because none were needed. No witnesses posted their own angle because there was nothing to witness. The building in the photo, once people compared it against street-level views, didn't even match the real Pentagon's structure; a floating lamp post and a strangely placed black post gave away that the scene had been assembled rather than photographed.
The same pattern showed up on a larger scale during coverage of the Iran war that began in early 2026. Fact-checkers at Poynter documented AI-generated footage claiming Iran had sunk the USS Abraham Lincoln, a claim easy to picture and, if true, enormous. It wasn't true. Analysts lined up the video against real photos of the vessel and found the hull markings didn't match. Other images circulating from the same period showed supposedly captured American soldiers with a Gemini watermark still faintly visible, the tool's own signature left behind in the picture it had made.
None of this took a studio or a team. A single convincing frame is now well within reach of one person with an ordinary laptop and a prompt, which is exactly why this category of fake has grown so quickly. You don't need to fabricate a whole news cycle: a burning building, a captured soldier, a sinking ship, one image can do the emotional work of an entire report, and it can be produced faster than any real crew could reach the scene it's pretending to show.
A striking image is not a report
A photo of a crowd, a disaster, an arrest hits you before you've had a chance to ask any questions of it, and that's precisely the design. Smoke, flashing lights, a dramatic angle: all of it is built to produce a feeling first and a fact second, if ever. But a single striking image, however convincing, is not itself evidence that the event it depicts occurred. It's evidence that a picture exists. Those are not the same claim, and treating them as the same claim is exactly how a fabricated explosion moves a stock index before lunch.
A picture of an event is not the event. It's a claim wearing a photograph.
The test that actually works
Big real events are witnessed by more than one camera, from more than one angle, by people with no reason to coordinate a story. A genuine explosion at a landmark building generates emergency dispatches, other people's phone footage, official statements, wire photos from more than one agency, all within minutes. A fabricated one generates exactly one image, endlessly reposted, often from an account with no track record of covering that beat. So the useful question isn't 'does this photo look real?' It's 'who else, independent of whoever posted this, is reporting the same thing?'
That's a search you can run in under a minute: take the core claim, drop the emotional framing, and look for it in a couple of other places you'd expect to cover it, official channels or established news organisations among them. For the Pentagon photo, the answer arrived fast: fire officials and the Pentagon itself said nothing had happened. For the sunk carrier, no navy confirmed a loss, no other footage matched, and the ship in question kept right on existing. Silence from everyone except the original poster is itself an answer, and it's often the fastest one you'll get.
It's worth being honest about the limit of this method too: a brand-new, genuinely important event can be minutes old and still real, with the wider coverage simply not caught up yet. The difference is what you're waiting for. A true event accumulates confirmation as the minutes pass: more footage, an official statement, a wire report. A fabricated one stays exactly as alone at minute sixty as it was at minute one, because there was never anything out there for other cameras to find.