Is TikTok's algorithm making eating disorders worse? What the 2024-2025 studies found

Is TikTok’s algorithm making eating disorders worse? What the 2024-2025 studies found

Concern about TikTok’s algorithm and eating disorders has moved from parenting forums into peer-reviewed research over the past two years. Several 2024 and 2025 studies have started asking a more precise question than “does TikTok cause eating disorders”: does the recommendation system that curates each person’s For You Page tend to serve more of this content to people who are already unwell, and if so, what does that actually do?

The tiktok algorithm eating disorders evidence base is still young, but it’s no longer just anecdote. Researchers have compared real TikTok feeds between people with and without eating disorders, tracked how engaging with appearance-focused videos changes what gets recommended next, and interviewed young people in recovery about what the experience feels like from the inside.

None of this shows that TikTok, on its own, causes an eating disorder to develop. Eating disorders have complex genetic, psychological and social causes, and no single app is a sole trigger. What the research does show is more specific: for people already vulnerable, the algorithm can amplify exposure to appearance, diet and pro-eating-disorder content in a way that plausibly makes things harder, not easier. This article walks through what was actually tested, what was found, and where the evidence still falls short.

What the Griffiths study actually did and found

The most direct piece of evidence comes from a 2024 study by Scott Griffiths and colleagues, published in the journal Body Image. Rather than asking people to describe their TikTok experience from memory, the researchers captured what over 90 participants were actually being shown: they collected roughly a million TikTok videos delivered to 42 people with a current or past eating disorder (about three-quarters had a history of anorexia nervosa) and 49 people with no eating disorder history, over one month of ordinary use.

The team then classified videos into categories relevant to eating disorder psychology: appearance-focused content, dieting content, exercise content, and explicitly pro-eating-disorder (“pro-ana”-type) material. The finding was not that everyone’s feed looked the same. Participants with eating disorders were, on average, shown meaningfully more videos in all four categories than the control group. And within the eating-disorder group itself, people with more severe symptoms had algorithms skewed even further toward this content.

That’s an important distinction. This isn’t a study of what TikTok shows a “typical” teenager. It’s evidence that the recommendation system responds to a user’s existing engagement patterns in a way that, for people already struggling, tends to concentrate rather than dilute exposure to risky content. A corrigendum was later published for this paper, correcting minor details rather than the headline sample or conclusions. The study remains observational: it shows association between illness severity and algorithm content, not proof that the algorithm alone worsened anyone’s condition.

The appearance-comparison mechanism: what the Dondzilo research adds

A second 2024 study, led by Ivana Dondzilo and published in the International Journal of Eating Disorders, helps explain a plausible mechanism behind the Griffiths findings: why engaging with this content might matter psychologically, not just statistically.

The study tracked how people’s engagement with appearance- and eating-related TikTok content related to two things over time: how much similar content the algorithm went on to recommend, and participants’ eating disorder symptoms. The pattern held together as a chain. Engaging with appearance and eating content predicted the algorithm serving up more of it. Being shown more of it predicted higher “upward appearance comparison”, the common habit of comparing your own body unfavourably to images you see. And more upward comparison predicted higher eating disorder symptom scores.

This matters for the causation question specifically because appearance comparison is a well-established, long-studied risk factor for disordered eating that predates social media entirely, it shows up in research on magazines and television going back decades. What’s new here isn’t the comparison mechanism itself; it’s evidence that an algorithm can create a feedback loop, where engagement narrows the feed, the narrowed feed increases comparison, and increased comparison correlates with worse symptoms. It’s a correlational design, so it can’t rule out that people with more symptoms simply seek out and engage with more of this content in the first place, likely some of both are true.

What recovering patients and young people describe

Alongside the quantitative studies, qualitative research has captured how this looks from the inside. A 2025 study in the Journal of Eating Disorders, led by Tamsin Parnell and colleagues at institutions including the University of Edinburgh, interviewed 17 UK-based young people aged 18–25 with current or past eating disorders about their experience of TikTok’s algorithm.

Three themes emerged. Participants described “view one, see more”, watching a single video and then finding themselves shown a steady stream of similar content, with a strong sense that they had little control over it. They described “morbid curiosity”, knowing, intellectually, how the algorithm worked and that certain content wasn’t good for them, but feeling a compulsion to keep watching anyway. And they described content drifting “from helpful to unhelpful” over time, where recovery-focused videos they’d sought out for support gradually gave way to more triggering material, because the algorithm doesn’t reliably distinguish a recovery account from a pro-eating-disorder one.

Related qualitative work has documented young people using “algospeak”, deliberately misspelled or coded words, to discuss eating disorder topics in ways that evade content moderation filters, a pattern that complicates platforms’ ability to detect and limit harmful content even when they are trying to. Several participants in these studies specifically described TikTok content as a relapse trigger during recovery. These are self-reports from a relatively small, self-selected group, so they can’t be generalised to all TikTok users, but they add texture that the quantitative data alone can’t: a lived sense of feeling pulled somewhere you didn’t choose to go.

What this evidence does and doesn’t prove

It’s worth being precise about the limits here, because overstating them would be its own kind of harm. Sample sizes in the core studies are small by public health standards, dozens of participants, not thousands, which limits how confidently findings generalise to the full TikTok user base. The Griffiths and Dondzilo studies are correlational: they show that certain patterns travel together, not that the algorithm is the sole or even primary cause of anyone’s eating disorder. People with more severe symptoms may both seek out and be served more of this content, and untangling which comes first is genuinely difficult with these designs.

A broader 2025 study in Pediatric Research, analysing survey data from over 220,000 adolescents across 41 countries, found that problematic social media use (not TikTok specifically) was linked to worse dietary patterns generally, lower fruit and vegetable intake, more sweets and sugary drinks, less regular breakfast, with effects varying by country and gender. This tells us social media use patterns relate to eating behaviour broadly, but it’s a different, blunter measure than eating disorder symptoms specifically, and it doesn’t isolate TikTok’s algorithm as a cause.

Taken together, the honest summary is: TikTok’s recommendation system appears able to amplify and concentrate eating-disorder-adjacent content for people already vulnerable, through a plausible comparison-based mechanism, and people in recovery describe this as a real relapse risk. That is meaningfully different from saying TikTok causes eating disorders outright, and the difference matters both for how seriously we take the research and for how it should be acted on.

The takeaway

The research on TikTok’s algorithm and eating disorders is still developing, but the emerging picture is consistent: for people already experiencing an eating disorder, the platform’s recommendation system can concentrate exposure to risky content in ways that plausibly interact with recovery and relapse. That’s different from proving TikTok causes eating disorders in the first place, and it’s a distinction worth holding onto rather than reaching for the more alarming version of the story.

If any of this feels close to home, for you or for someone you care about, please don’t wait to reach out. Beat, the UK’s eating disorder charity, runs a free, confidential helpline staffed by trained advisers.

This article summarises published research for general information and isn’t medical advice. If you’re worried about your own or someone else’s eating, please speak to a GP or a service like Beat.

FAQs

Does TikTok’s algorithm cause eating disorders?

The current evidence doesn’t show that TikTok’s algorithm causes eating disorders to develop from nothing. What studies do show is that for people who already have an eating disorder, the algorithm can amplify exposure to appearance, diet and pro-eating-disorder content, which may interact with existing symptoms and recovery.

What did the main 2024 TikTok eating disorder study actually find?

A study by Griffiths and colleagues, published in Body Image, compared real TikTok feeds from 42 people with eating disorders and 49 people without over one month. It found participants with eating disorders were shown significantly more appearance, diet, exercise and pro-eating-disorder content, and that content skewed further toward this material as symptom severity increased.

Is this research proof or just correlation?

Mostly correlation. The core studies show that certain patterns, engagement, algorithm output, and symptoms, travel together over time, which is meaningful evidence, but they can’t fully rule out that people with more severe symptoms simply seek out more of this content themselves. Sample sizes are also small, which limits how far the findings generalise.

What is “algospeak” and why does it matter for eating disorders?

Algospeak refers to deliberately misspelled or coded words people use to discuss sensitive topics, including eating disorders, in ways designed to evade a platform’s content moderation. Researchers have noted its use in eating disorder communities on TikTok, which complicates efforts to detect and limit harmful content even when platforms are trying to.

Can TikTok trigger a relapse for someone in eating disorder recovery?

Some people in qualitative studies have described exactly this, TikTok content shifting from supportive recovery material to more triggering content over time, and experiencing this as a relapse risk. This is based on self-reported experience from relatively small groups, not a large-scale trial, but it’s a consistent theme across more than one study.

Does this research apply to all social media, or just TikTok?

Most of the specific algorithm-comparison research so far has focused on TikTok because of its heavily algorithm-driven feed. A broader 2025 study across 41 countries found problematic social media use in general (not tied to one platform) was associated with worse dietary patterns in teenagers, suggesting the underlying concern isn’t unique to any single app.

What should parents or carers do with this information?

Treat it as a reason for open conversation rather than panic. Understanding that algorithms can amplify existing struggles is useful context if you’re worried about a young person’s TikTok use or eating patterns, but a specific concern about someone’s health is best raised with a GP rather than managed alone.

Where can I get help for an eating disorder in the UK?

Beat, the UK’s eating disorder charity, runs a free confidential Helpline on 0808 801 0677 (3pm–8pm, Monday to Friday), plus a Youthline, Studentline and webchat support at beateatingdisorders.org.uk. Your GP can also refer you to NHS eating disorder services. In a crisis, call 111 or go to A&E.

Is “pro-ana” content still on TikTok?

Explicitly pro-eating-disorder content is against TikTok’s community guidelines and the platform does moderate it, but researchers have found it still surfaces, partly because of algospeak and coded language designed to evade detection. Studies like Griffiths et al.’s specifically measured how much of this content reached vulnerable users despite moderation efforts.

Does watching recovery content on TikTok help or hurt?

It can do either. Some participants in qualitative research found recovery-focused accounts genuinely supportive, but also described the algorithm gradually blurring the line between recovery and pro-eating-disorder content, since the system isn’t reliably able to tell the two apart based on similar visual and topic cues.