She’s not imagining it. TikTok’s algorithm favours mental health content over many other topics, including politics, cats and Taylor Swift, according to a Washington Post analysis of nearly 900 US TikTok users who shared their viewing histories. The analysis found that mental health content is “stickier” than many other videos: it’s easier to spawn more of it after watching with a video, and harder to get it out of your feed afterward.
“It felt like a rabbit hole to me because you kept going down deeper and deeper,” Russell said.
TikTok uses an algorithm to select a video and gives users two main options: watch it or skip past to something else. Along the way, the app learns what a user like Russell likes and dislikes, based on her watching and skipping behaviour. It takes skipping past 1.3 videos, on average, to undo the effect of watching one full video about cats or politics, The Post analysis found. For mental health, it takes 2.2 skips – meaning users must work harder to get it out of their feeds.
TikTok spokesperson Mahsau Cullinane criticised The Post’s methodology as incomplete and said it doesn’t “reflect the reality of how our recommendation system works”.
This finding comes amid a broader debate on the role of algorithms and influencers in Americans’ understanding of mental health. Content about mental illness and neurological differences is extremely popular across social media apps, with about as many TikTok posts using the hashtag #mentalhealth as those that mention #sports, according to data from analytics firm Sprout Social. Mental health content on TikTok deals with not just conditions like depression or anxiety, but also living with a neurological type such as ADHD or autism.
People are turning to social media for health information as Americans face a shortage of mental health professionals, barriers to accessing and paying for care, and lingering stigma. Information from social media helps underserved and underdiagnosed populations better understand themselves, many users say. What happens next, however, is rarely examined.
Over the period that The Post examined Russell’s TikTok data, about one in 11 videos on her feed were mental-health-related. Russell, who spent more than an hour watching videos on many days, said the more she scrolled, the more often she saw videos from non-professionals that seemed designed to get a reaction rather than educate.
Efforts to evaluate mental health content on TikTok support Russell’s impression. Anthony Yeung, a psychiatrist and University of British Columbia researcher, ran a study examining 100 top TikTok videos about ADHD and found that some were helpful, but about half were misleading. (Videos about creators’ personal experiences weren’t classified as misleading.) Other reviews of TikTok content about ADHD and autism by mental health practitioners have found similar results.
“The algorithm says, ‘Well, you like this video about ADHD, even though it’s misleading, let’s give you another video,’” Yeung said. “And it becomes this very vicious feedback loop of misinformation.”
The phenomenon is having a profound effect on real-world mental health treatment, clinicians say. Yeung said he deals with “two visions of what ADHD is”: the one discussed on social media and the one he sees among actual patients. On TikTok, ADHD content often paints with a broad brush, portraying common quirks or struggles as not just personal experiences but diagnostic criteria for the condition.
One popular ADHD account, @lifeactuator, regularly earns views in the millions with titles like “What ADHD feels like” and “Things people with ADHD do despite knowing better”. One widely watched video with the caption “if the world was made for ADHD” depicts a Costco store with ADHD shoppers being chased around by store employees to stop them from making impulse purchases.
Eric Whittington, the Arizona-based creator behind @lifeactuator, said that because of the constraints of short-form video, he’s not able to include all the information viewers might need to understand what, if anything, his videos reflect about ADHD as an actual medical condition. Taken individually, his videos probably apply to a broad swath of the population, he said – not just people with ADHD.
“When you only have a minute to work with, it’s hard to add disclaimers on the content saying, ‘Yes, everybody experiences this from time to time, but if it happens all the time, you may have ADHD,’ ” he said.
Rana Coniglio, an Arizona-based therapist who works primarily with Gen Z clients, said they often arrive at her practice already attached to a diagnosis they found on TikTok. Sometimes, that attachment makes it harder to accurately diagnose or make a treatment plan that could improve that person’s symptoms.
“I have had people come to me and say, ‘Hey, I saw this video on TikTok and it’s actually the reason that I’m seeking therapy because it made me think I actually do need help,’ and there are benefits to that,” she said. “But I think the majority of people see a diagnosis, take it and run with it.”
High volume, low quality
For Ace Bannon, a 19-year-old in Utah, the more he watched, the darker the content became.
Bannon first got curious about autism and its characteristics after learning that many of his best friends – people he’d met on a Discord server – were autistic. He started watching TikTok videos, with content about autism taking up a growing chunk of his feed. Then, TikTok served him video after video of autistic adults discussing the trauma they endured as children, Bannon said. Before long, he wanted his old algorithm back.
“Because you’re interested, it starts recommending more of those videos and it makes you fall into these rabbit holes that you just want to get out of after a while, but you can’t.”
Sometimes this experience actually exacerbates existing mental health problems, some users say. Kailey Stephen-Lane, 30, said she had to temporarily stop using the app because spending time on TikTok was worsening the symptoms of her obsessive compulsive disorder. While her real-life therapist was helping her sit with fears and insecurities without fixating, TikTok was “bombarding” her with videos about the very symptoms that made her so anxious, she said.
“The TikToks that I’ve been getting are not helpful to my recovery,” she said. “They lead me down a lot of spirals, and me just clicking ‘not interested’ doesn’t seem to work anymore.”
TikTok provides a high-level description of some of the data its algorithm uses but few details. That makes it difficult to know why mental health videos are stickier than other topics, says Stevie Chancellor, an engineering professor at the University of Minnesota who studies AI and its risks, and whose research found that the algorithm creates a “runaway train” of mental health content.
But the app’s business incentives offer some clues, Chancellor says. Maybe users who see a lot of mental health videos spend longer on the platform or are more likely to spend money down the line, she said. Maybe the effect is completely unintentional, an example of a black-box algorithm optimising for what it thinks users want.
“Watching [mental health] content might lead to other behaviours that are valuable on the platform,” Chancellor said.
The topic may become sticky because it’s one “that a user only wants to engage with sometimes,” said Laura Edelson, a computer science professor at Northeastern University who collaborated with The Post in a parallel TikTok research effort.
Cullinane, the TikTok spokesperson, said the company is “transparent” about how its feed works.
For TikTok users, adjusting the type of content that shows up on their feeds can be hard. It’s not always clear when engaging with a certain video would spawn something undesirable: even watching clips about romantic relationships made a user more likely to encounter mental health content, The Post’s analysis found. TikTok has gradually added options that could help users tailor their feeds, such as clicking a “not interested” button, blocking videos with certain keywords or resetting their algorithms from scratch. A new “Manage Topics” menu lets users adjust the prevalence of 12 specific topics on their For You page – but mental health isn’t one of them.
As for Russell, she is glad for the journey toward an ADHD diagnosis because of TikTok. She just wishes her favourite type of content – lighthearted cat videos – got the same treatment from the app’s algorithm.
“I want like 10 to 20 per cent cute cat videos, probably even like 30 per cent,” she said. “But those disappear really quickly.”
Hundreds of TikTok users in the United States sent their watch history data to The Washington Post. We downloaded the collective 14.8 million videos they’d been shown and then sorted them into topics, based on keywords in the transcripts and on-screen text. The Post calculated the stickiness of each topic by computing the difference between the number of topical mental health videos each user had been shown in the previous 50 videos and how many they saw in the next 50. We averaged this for all videos, aggregated by whether the user watched at least 90% of the video, or skipped it.