Ever since the explosion of AI image generators a few years ago there's been a fear that the technology would eventually be able to generate images so realistic that it would become impossible to know what's real online. The Emily Hart AI hoax was a sign that we're reaching that point. AI photos and videos are being used to make fake news and fake people that are scarily believable.
But it turns out that dubious actors don't necessarily need flawless photorealism to mislead people. A lot of AI imagery still has an uncanny valley effect, and it turns out that this weird visual quality can be useful for anyone pumping out misleading ads or outright scams. Those ubiquitous tai chi walking ads show why.
If you use YouTube, you'll probably have seen the adverts for supposed tai chi walking programmes claiming that within weeks an 80-year-old man who has never done any exercise can obtain the lean physique and six-pack abs of a fictional AI-generated guru who spends most of his time shirtless.
Thousands of variations of the ads have appeared across social media platforms, mainly targeted at over 40s interested in light fitness, yoga, or “health hacks.”
With the ads all over YouTube, several creators on the platform began looking into them. In the video above, Josh Brett, who's making of a series of short documentaries looking into aspects of the health and fitness industry, reveals how they work.
The ads' claims are so absurd that most people laugh and move on, assuming the product to be a scam. The AI-generated imagery makes it seem even more obviously fake. And it seems that might be the point.
The ads appear to promote an app offering a generic AI-generated strength training and diet plan with no connection to tai chi. Several reviews on sites like Trustpilot claim the app is a scam, with subscriptions costing more than advertised, often padded out with unexpected and expensive add-ons that are hard to cancel once you hand over your credit card details.
The use of bad AI video essentially provides a visual enhancement to the classic Nigerian Prince scams. It filters out the most skeptical viewers so the advertiser doesn't waste time and money on them. Those who end up clicking through are already more likely to believe in the implausible program, more likely to accidentally click on the windows offering expensive add ons and ultimately less likely to realise they've been tricked or to try to do anything about it.
Of course, generative AI has entered mainstream advertising too, as our review of 2026 World Cup adverts shows. It seems the backlash against the McDonald's and Coca-Cola Christmas ads wasn't enough to put brands off the tech completely.
