Deepfake doctors are AI-generated videos or audio that copy the face and voice of real, trusted physicians to sell supplements, “cures” or scams. The BMJ reported in 2024 that well-known UK TV doctors were being faked to promote products for blood pressure and diabetes. Be suspicious of any doctor video selling something, check the claim with an independent source, and never stop prescribed treatment because of an online video.
The video looks right. A familiar face, a doctor you have seen on morning television for years, sits in what looks like a studio. The voice is his. He explains that a natural remedy lowers blood sugar better than the medicine your GP prescribed, and that pharmaceutical companies do not want you to know. There is a link underneath.
He never said any of it. The face was copied, the voice was synthesised and the link leads to a shop that did not exist a month ago.
In July 2024 The BMJ published an investigation showing that some of the UK’s best-known TV doctors, including Hilary Jones, Rangan Chatterjee and the late Michael Mosley, were being “deepfaked” in videos promoting products for conditions such as high blood pressure and diabetes, including supplements like CBD gummies. John Cormack, a retired doctor, worked with the journal to gauge the scale of the problem, finding many examples on Facebook and Instagram. Meta told The BMJ it would investigate the examples and does not permit content designed to deceive or defraud.
The argument of this piece is that deepfake doctors are not just another flavour of online misinformation. They attack the thing medicine runs on: the trust that attaches to a known clinician’s face and voice. That makes them a professional problem for clinicians, not only a consumer protection problem for platforms.
What is a deepfake doctor?
A deepfake uses AI to generate or alter video and audio so that a real person appears to say things they never said. The technology has become cheap and easy enough that a convincing short clip can be made from publicly available footage of anyone who appears regularly on television, podcasts or social media.
Doctors are attractive targets for an obvious reason. Surveys in many countries consistently rank doctors among the most trusted professions. A supplement advert from an anonymous influencer invites scepticism. The same claim from a doctor viewers feel they know can bypass it.
The typical pattern is familiar from older scams, just with a better mask. A real clinician’s likeness, an urgent health claim, an attack on conventional treatment, and a link to buy something.
How is AI making health misinformation worse?
Deepfakes are the most visible part, but AI is changing health misinformation in three wider ways.
Volume. Language models can generate endless variations of persuasive text, fake testimonials and “research summaries” at almost no cost. Moderation systems that rely on spotting repeated content struggle with material that is never quite the same twice.
Authority. Misinformation used to look amateur. AI output is fluent, formatted and confident, and it can mimic the tone of a medical journal or a hospital website.
Weaponised chatbots. In June 2025 researchers led by Natansh Modi at the University of South Australia published a study in Annals of Internal Medicine testing whether five widely used AI models could be turned into health disinformation chatbots using hidden instructions of the kind any developer can set. They instructed each model to give false answers to health questions in a formal, scientific and authoritative tone. Across 100 health questions, 88% of the responses were disinformation. Four of the five models produced disinformation in every one of their responses; the fifth did so in 40%. The instructions were invisible to anyone chatting with the finished bot, which would simply appear to be a knowledgeable, polite health assistant.
This matters because the public increasingly asks chatbots health questions. A chatbot embedded in a supplement seller’s website, or dressed up as a “health assistant,” may look neutral while being configured to mislead.
Why deepfake doctors are a trust crisis for clinicians
The harm is not limited to people who buy the product.
When a real doctor’s face is used to sell a fake cure, it damages that doctor’s reputation. Patients may lose trust not only in the doctor but in medical advice in general, once they learn that even a familiar face on screen can be fake. The rational response to “you can’t trust what you see” is often to trust less, including accurate public health messages.
This deepens an existing problem. We argued in an earlier piece on vaccine hesitancy that trust is built through relationships and access more than through information alone. Deepfakes exploit precisely those relationships, borrowing the credibility of clinicians whom the public has come to know.
There is also a direct clinical risk. Many deepfake adverts target chronic conditions such as diabetes and high blood pressure, where the promised “natural cure” competes with effective medicines. Someone who stops insulin or blood pressure tablets in favour of a supplement can come to serious harm.
A worked example: checking a suspicious video
This example is hypothetical.
Grace, 63, has type 2 diabetes. She sees a video on social media in which a well-known doctor says a herbal supplement “reverses diabetes in 30 days” and that metformin “does more harm than good.” The doctor looks and sounds real. The video has thousands of likes and many positive comments.
Before buying, she runs through a few checks.
She looks at the account. It is not the doctor’s verified account; it was created recently and posts only product adverts. She searches the doctor’s name with the word “deepfake” and finds that he has publicly warned that his likeness is being misused. She notices the video urges her to buy quickly, with a countdown timer. And she remembers that no legitimate doctor would tell a stranger online to stop prescribed medicine.
She does not buy it. At her next appointment she asks her GP about supplements, and they talk through what does and does not help with blood sugar.
The checks took five minutes. None of them needed technical skill to spot visual glitches, which is fortunate, because as the technology improves those glitches are disappearing.
How can you spot a deepfake doctor or AI health scam?
Technical tells such as odd blinking, mismatched lip movement or strange lighting can help, but they are becoming unreliable. It is safer to judge the claim and the context rather than the pixels. Warning signs include:
- A doctor selling a product directly, especially a supplement, with a buy link
- Miracle claims: cures, reversal of chronic disease, results in days
- Attacks on standard treatment, such as telling people to stop prescribed medicines
- Urgency and secrecy: limited-time offers, claims that “they” are hiding the truth
- An unfamiliar or new account rather than the doctor’s own verified channels
- Comment sections full of glowing testimonials that read alike
If you are unsure, check the doctor’s official website or verified accounts, look for news reports of misuse, or ask a pharmacist or your own clinician. Never stop or change a prescribed medicine based on an online video.
The strongest counterargument: isn’t this just old quackery in new clothes?
It is a fair challenge. Snake oil is older than television. Fake endorsements have been used in adverts for decades, and health misinformation spread widely long before generative AI. Some researchers argue that the most harmful misinformation still comes from real people with large followings rather than from synthetic ones, and that focusing on deepfakes distracts from that.
There is truth in this. The underlying scam is old, and much misinformation, as we noted in our piece on social media and teen mental health, is a platform and incentive problem as much as a technology one.
But two things are genuinely new. The cost of producing a convincing fake endorsement from a specific, trusted person has collapsed. And the target is no longer “a doctor” in the abstract but your doctor, the one you recognise. That combination scales a very old con to a level older safeguards were not designed for.
What this means for clinicians, platforms and regulators
For clinicians. Assume that if you have a public profile, your likeness can be copied. Maintain a clear, verified presence so patients know where your real advice lives, and say publicly that you do not sell products through adverts. Ask patients directly about supplements and online advice; many will not volunteer it. Professional bodies can help by running shared reporting routes and supporting members whose identities are misused.
For platforms. Adverts that use a real doctor’s likeness to sell health products should face stronger verification before they run, not just takedown after complaints. Faster removal when a doctor reports impersonation would reduce the harm window.
For regulators and policymakers. Existing rules on advertising, consumer protection and impersonation apply, but enforcement is slow compared with how quickly fakes spread. Clear labelling requirements for synthetic media, and accountability for platforms that profit from fraudulent health adverts, are live policy questions in many countries.
For patients. Be sceptical of any doctor selling something online. Your own clinician is the right person to ask about supplements, and no genuine treatment requires you to buy it within the next ten minutes.
Frequently asked questions
What is a deepfake doctor?
A deepfake doctor is an AI-generated or AI-altered video or audio clip that makes a real physician appear to say things they never said, usually to sell supplements or “cures.” The BMJ reported in 2024 that well-known UK TV doctors were being faked this way on social media to promote products for conditions such as diabetes and high blood pressure.
How can I tell if a doctor video is fake?
Visual glitches are becoming harder to spot, so focus on the context. Be suspicious if a doctor is selling a product directly, promising cures, attacking standard treatment, creating urgency, or posting from an unfamiliar account. Check the doctor’s official website or verified channels and ask your own clinician or pharmacist before acting.
Can AI chatbots spread health misinformation?
Yes. A 2025 study in Annals of Internal Medicine found that five widely used AI models could be configured with hidden instructions to give false health answers, and 88% of their responses to health questions were disinformation. Chatbots are also sometimes simply wrong. Verify important health information with a clinician or reputable health organisation.
What should I do if I see a deepfake doctor advert?
Do not buy the product or click through. Report the advert to the platform as a scam or impersonation. If you know the doctor or their organisation, let them know. If you have already bought something or shared payment details, contact your bank and report the fraud to the relevant consumer protection or fraud authority.
Is it dangerous to follow health advice from social media videos?
It can be. Videos promoting “natural cures” often discourage proven treatment for conditions such as diabetes and high blood pressure, and stopping prescribed medicine can cause serious harm. Treat social media health advice as a prompt for a question to your clinician, never as a reason to change your treatment on your own.
The fake doctor is selling gummies. What he is really spending is the trust real doctors spent careers building. Clinicians who stay silent about it will find that trust harder to borrow back.
Have you, or a patient, encountered a deepfake doctor online? How did you realise it was fake?
