Half of what just felt like a diagnosis was, more likely than not, a guess dressed up as certainty.
Your thumb finds the app before your eyes have properly opened. It is twenty past seven, the curtains are still drawn, and before you have even decided to get up, you are three videos deep into someone explaining, in fifty-nine confident seconds, the six signs that you have ADHD. Two of the six describe your Tuesday exactly. A small, warm jolt goes through you — the sense of being caught out, recognised, finally explained. You screenshot it to send to a friend. What you do not yet know, lying there in the grey morning light, is that the person on your screen has, roughly, a coin’s toss chance of being right.
That recognition is real. The odds behind it are not good.
Perhaps that has already happened to you more than once. Perhaps you are reading this while you wait for an assessment — one of an estimated 735,000 people in England alone with an open ADHD referral as of the end of last year, watching a wait that your own local NHS service has admitted, apologetically, could run from a few months to several years depending on nothing more meaningful than your postcode. Perhaps you were diagnosed decades ago and are only now, watching strangers narrate your own childhood back to you in a kitchen or a car, understanding what that diagnosis actually meant. Or perhaps you are a parent, phone propped against the kettle, wondering whether the meltdown your child had over a sock seam this morning is something to mention at parents’ evening, or something with an actual name. Whichever of those is you, a piece of research published this year gives you something steadier to hold than a stranger’s confident voiceover — not a verdict on TikTok, but a number.
What happened when researchers read five thousand posts
Dr Eleanor Chatburn and Dr Alice Carter, working out of the University of East Anglia’s Norwich Medical School, did something almost nobody watching this content ever does: they sat down and checked it against the evidence. Their systematic review, The Quality of Mental Health and Neurodivergence-Related Information on Social Media, published in the Journal of Social Media Research in 2026, worked through more than five thousand posts across five platforms — TikTok, YouTube, Facebook, Instagram and X — covering autism, ADHD, schizophrenia, bipolar disorder, depression, eating disorders, OCD, anxiety and phobias. Every claim was measured against clinical guidance and diagnostic criteria, not against how convincing it sounded.
The results were not close. Across the platforms studied, misinformation rates reached as high as fifty-six per cent, and TikTok came out worst of all. Worst of all, on the app you are most likely to be watching. Fifty-two per cent of the most-viewed ADHD videos on the platform contained claims that were inaccurate or unsupported by evidence. Forty-one per cent of autism content fared the same way. By contrast, YouTube averaged twenty-two per cent misinformation, Facebook sat just under fifteen per cent, and YouTube Kids — a heavily moderated, child-facing corner of the same company — scored zero per cent misinformation on anxiety and depression content and just 8.9 per cent on ADHD.
The most telling comparison in the whole study, though, is not platform against platform. It is creator against creator, on the very same app. When Chatburn and Carter separated TikTok’s ADHD videos by who made them, professionally produced content — filmed by psychiatrists, psychologists, or clinicians speaking within their own field — carried a misinformation rate of just three per cent. Content made by people without that training, however sincere, however many likes it had collected, carried a misinformation rate of fifty-five per cent. The platform is not uniformly unreliable. It is unreliable in a very specific, very findable place: wherever a confident amateur has replaced a qualified professional, and the algorithm has not noticed the difference. Does that distinction — professional voice against enthusiastic amateur — match how you actually choose what to watch at half past eleven, scrolling with the sound down low so you don’t wake anyone?
Why the algorithm does not care if it is true
It is worth being precise about what is actually happening to your feed, because it is not simply that some people online get things wrong. Misinformation has always existed. What TikTok adds is a distribution system that was never built to sort truth from noise in the first place — it was built to sort engagement from noise, and accuracy and engagement are only loosely related. As the psychiatrist Dr Darren O’Reilly put it in response to the UEA findings, “TikTok does not reward being right. It rewards being relatable, confident, and easy to share.”
A fifty-nine-second video that says “you might have ADHD if you procrastinate, then panic, then somehow finish everything at 2 a.m.” will always outperform a psychiatrist carefully explaining that ADHD is diagnosed through pervasiveness across settings, onset in childhood, and functional impairment measured over time — because the first version is instantly, viscerally relatable, and the second one requires patience you probably do not have at half past eleven. The app’s recommendation system learns what keeps you watching, not what is clinically sound, and it has no mechanism for telling the two apart. It is not choosing to mislead you. It was never built to check. Once you pause on one video about racing thoughts or rejection sensitivity, the system reads that pause as a signal, not a symptom, and quietly rearranges your feed around it.
The echo chamber you built without meaning to
Within days, what felt like your own private realisation starts arriving from a dozen different creators, in a dozen different accents, filmed in bedrooms that look a little like yours. It can feel like confirmation, the way a rumour starts to feel true once you have heard it from three separate people, forgetting that all three heard it from the same original source. It is not confirmation. It is the same recommendation engine, showing you variations on the video you already lingered over, because lingering is the only language it speaks. You are not gathering evidence. You are being shown more of what already caught you.
Why a stranger’s video can feel like it is reading your diary
There is a second layer to this, and it sits inside your own head rather than inside the app. Psychologists call it the Barnum effect, named after the showman P.T. Barnum: the tendency to accept a vague, near-universal description as uniquely, personally true. “You zone out mid-conversation. You start ten projects and finish two. You feel everything more than everyone around you seems to.” Read those sentences again, slowly, as if a stranger had written them just for you. They describe most human beings on most difficult weeks. But heard at speed, in a video edited for maximum emotional impact, they do not feel general. They feel like being seen for the first time.
Underneath that is a harder truth the researchers kept returning to: nearly every clinical symptom is simply a normal human trait at an elevated volume. Everybody procrastinates sometimes. Everybody loses their keys, zones out in a meeting, feels crushed by a small rejection now and then. What separates a trait from a disorder is not the trait itself but four unglamorous, unquotable words that rarely survive a fifty-nine-second edit: severity, persistence, pervasiveness, and impairment. Does it happen most days, in most areas of your life, badly enough to actually get in the way? A fifteen-second clip cannot ask you that question. Only a proper assessment can. The recognition you feel scrolling at midnight is real. The diagnosis is not — not yet, not from that alone.
The part of this story that is not bad news
It would be unfair to leave you with only the fifty-two per cent, because that is not the whole picture, and you deserve the rest of it too. That same three per cent figure for professional TikTok content is not a rounding error — it is proof that good, accurate, genuinely useful mental health information exists on this exact platform, made by people qualified to give it, and it is being watched by millions who would otherwise have no access to it at all. For years, ADHD and autism in adults — particularly in women, who were routinely overlooked by diagnostic criteria built and tested largely around how both conditions present in boys — went unrecognised for entire lifetimes, dismissed instead as anxiety, or shyness, or simply being “a bit much.” If you are one of the 735,000 people currently waiting for an NHS assessment, you may be on that list only because a video, made by an actual clinician, gave you a word for something a GP had spent years quietly missing. Social media, whatever its faults, has put language and recognition within reach of people the traditional system failed for a generation. The problem this research identifies is not that people are talking about mental health on TikTok. It is that, right now, the app cannot tell you which fifty-nine seconds to trust — and, without a little help, neither can you.
Before you take this to your GP
Here I want to slow down, because hope without honesty helps no one, and getting this wrong could cost you more than it saves you. Before you act on anything above, you should know the shape of what this research cannot tell you.
It measured the accuracy of claims made in videos, not what happened afterwards to the people who watched them — so it cannot tell you whether a stranger’s video will send you towards the right appointment, talk you out of one you genuinely need, or simply leave you carrying a screenshot around for months, half-convinced, never quite sure enough to book it. It looked at the most-viewed content, which means a quieter, more careful video watched four hundred times rather than four million was never counted, and yours may have been exactly that kind of video. And a review like this is a photograph, not a live feed: by the time you read this, the creators, the moderation policies and the algorithm itself will all have shifted a little, in directions nobody can promise are in your favour.
None of that undermines the fifty-two per cent. It does mean that number describes a pattern in what you are currently being shown, not a rule for what will happen to you personally — and it is not a substitute for an actual clinical opinion in either direction, not the video’s, and, for what it is worth, not this article’s either. A screen cannot assess you. A person can.
How to watch without losing your bearings
None of this means closing the app for good, and it would be strange advice from someone asking you to keep reading a screen right now. It means watching differently. A few small, practical habits can do most of the work.
- Check who is talking before you check what they are saying. Look at the bio. A psychiatrist, clinical psychologist, or registered ADHD/autism specialist speaking within their own field carried a misinformation rate of three per cent in this research; an enthusiastic non-professional carried one of fifty-five per cent. That single check does more than almost anything else on this list.
- Listen for the four boring words. Severity, persistence, pervasiveness, impairment. A video that mentions how often, how long, how widely, and how much it actually disrupts your life is doing something closer to clinical thinking than one that simply lists relatable traits.
- Treat a video as a prompt, not a verdict. “This made me want to look into it properly” is a healthy reaction. “This told me what I have” is not one a fifty-nine-second clip is equipped to deliver.
- Take the recognition to a real conversation. In the UK, that usually starts with your GP, who can refer you into an NHS assessment pathway, or with a charity such as the ADHD Foundation or the National Autistic Society, both of which offer screening information built by clinicians rather than algorithms.
- Notice when your feed has narrowed. If every video you are shown now seems to be about the same condition, that is the recommendation system responding to your attention, not the universe confirming your suspicions. It is worth naming that shift to yourself when you spot it.
- If you are watching on behalf of a child, write down what you actually observe. Dates, settings, specific behaviour — that record will be more useful to a paediatrician or school SENCo (the staff member who coordinates support for additional needs) than any video, however precisely it seemed to describe your morning.
Where this leaves you
You do not have to delete the app to protect your mind from it, and you do not have to distrust every video you watch from tonight onward, either. You simply know something now that most people scrolling beside you do not: that the difference between being informed and being misled on that screen very often comes down to one detail you can check in about four seconds, and a handful of unglamorous words — severity, persistence, pervasiveness, impairment — that no algorithm has yet learned to reward. Carry those with you the next time a stranger tells you, with total confidence, exactly who you are.
Source: Chatburn, E. & Carter, A., “The Quality of Mental Health and Neurodivergence-Related Information on Social Media: A Systematic Review,” Journal of Social Media Research, University of East Anglia (Norwich Medical School), 2026.
Leave a comment