How-to·September 19, 2026, 10:00

Fake AI image of Trump fooled millions online

AI-generated and checked against the sources listed below.

An AI-generated image of the US president kissing a staffer was shared millions of times before it turned out to be fake. The case shows how hard it is to rely on AI tools to detect fake images.

AI-generated image

Last week, an image circulated on social media that purportedly showed US President Donald Trump kissing his staffer Natalie Harp on a golf course. The image was viewed more than 20 million times on the social network X, and the rumors flew.

The White House quickly dismissed the image as fake. Fact-checkers at outlets including Snopes could not find any credible source or original version of the image, and neither news agencies nor the White House's own photographers had taken it.

There were several signs that the image was made with AI. The sharpest version showed, among other things, a hand with six fingers, and a digital forensics expert found an invisible watermark called SynthID in the image. Watermarks of this kind are used to label content that has been created or edited with AI tools.

What's interesting is that several well-known AI chatbots were asked to assess whether the image was real, and they performed poorly. One chatbot even praised the image's natural depth of field and called the hands anatomically correct, while others gave vague answers. None of them noticed the most obvious flaw: the six fingers.

The story is a reminder that you shouldn't blindly trust AI tools when checking whether an image is real. Experts instead recommend looking at where the image comes from, whether recognized news agencies have published it, and whether details like hands, teeth and background make sense.

For Danish workplaces and ordinary users, the point is simple: AI images keep getting better, and they can spread quickly before anyone has a chance to check them. Don't share surprising images until you've seen whether a credible source has confirmed them. And remember that even advanced AI tools can be badly wrong when it comes to exposing their own kind.

Sources

More on this topic

Get the week's AI news in your inbox

Choose your level, topics and length. One email a week, unsubscribe at any time.

Subscribe to Promptly Newsletter
PromptlyNewsletterRSSLog in

The news on aijour is AI-generated and checked against the cited sources.