The advice arrives in the confident cadence of a physician. Microwaving food in plastic gives you cancer, a spoonful of Coca-Cola mixed with onion will ease your pain, and a supplement called “Hyalethinap Plus Pro Max” will fix your hair, nails, and joints in one go. The doctor delivering it does not exist.

A study published this month by Hallam, a UK digital marketing agency, found that AI-generated content appeared in 40% of the most-viewed health videos on TikTok, rising to 84% of the clips surfaced by searches for “health tips”. The researchers analysed 1,198 videos, and the top AI-generated ones averaged 2.5 million views apiece.

The findings land on a platform already thick with synthetic media. Nearly 60% of the videos TikTok shows new users are AI slop, a separate study found earlier this year, and dubious claims that AI outperforms doctors have already prompted warnings that patients could be misled at scale.

The doctors quoted in the research were blunt about the stakes. “Platforms like TikTok must do more to stop the rise of dangerous fake medical advice,” said Dr Emma Runswick, deputy chair of the British Medical Association council.

Professor Frankie Swords, NHS England’s national medical director, said many clinicians “are reporting an increasing number of patients who feel worried and confused due to false advice”.

Alex Ruani, a health misinformation researcher at University College London, was sharper still, describing the phenomenon as “industrialised exploitation of trust”.

The videos follow a recognisable template. A polished avatar, built with off-the-shelf tools, lip-syncs a script that blends plausible wellness tips with disproven myths and a pitch for a product that often does not appear to exist.

Hallam found no online trace of “Hyalethinap Plus Pro Max”, the anti-ageing remedy touted in one widely shared clip. Other clips recycle familiar cancer myths about microwaving food in plastic, deodorants, and sleeping near a phone.

This is not the first alarm. In February, the non-profit Science Feedback identified 18 such accounts within hours, several with hundreds of thousands of followers and videos reaching 14 million views.

The avatars claimed that turmeric lowers high blood sugar and that raw cacao and beetroot “unclog your arteries”, while pushing unproven supplements. Some cloned the faces of real physicians without consent, the kind of fakery that has spawned a small industry in deepfake detection.

The scale has unsettled the medical establishment on both sides of the Atlantic. John Whyte, chief executive of the American Medical Association, warned in February that scammers were cloning “the faces and voices of real, credentialed physicians” and called for physician identity to be treated as a protected professional asset.

He noted that one doctor’s own mother had been taken in by a fabricated video featuring her daughter, a measure of how convincing the clips have become. The same tools that promise gains in AI-assisted healthcare are now being turned against the patients they were meant to serve.

TikTok disputed the research. It said the findings came “from a marketing agency with commercial interests” and were “not an accurate reflection of our platform”, adding that it removes harmful health misinformation and partners with the World Health Organization and the NHS.

Its rules already prohibit both impersonation and misleading AI-generated content, and Science Feedback’s investigation noted that the company takes down most violating material proactively. The gap is between the policy and the feed.

The detection tools are improving, but the economics run the other way. One video app alone reportedly generates around 3 million clips a month, and accounts pulled for impersonation have been seen reappearing within the hour. The doctor may not be real, but the supplement sale is.

Get the TNW newsletter

Get the most important tech news in your inbox each week.