Feeds, filters, like counts and endless scroll hit women and girls hardest. Verified global evidence, regulation in 2026 and what platforms must change.
Social media now counts 5.81 billion user identities, about 70% of the world's population. Women and girls do not use it more than men, yet they carry a larger share of its health cost. The cause is not use as such. It is a set of design decisions: ranking for attention, editing faces by default, scoring approval in public and removing every reason to stop.
Those decisions are neither neutral nor irreversible. Courts and regulators on five continents now treat addictive design as a product defect. In March 2026 a Los Angeles jury found Meta and YouTube liable for harm to a young woman who started on their platforms as a child. In August 2026 a US federal court approved a settlement under which Meta pays about $12.2 billion and redesigns its teen accounts.
The health sector must stop treating this as a parenting problem. It is a women's health problem, and the fix is in the design.
The scale of the problem
In every market with reliable data, girls report more harm than boys from the same platforms.
Women make up 45.6% of the global social media audience, and the split varies by platform. Women are 65.5% of Pinterest's ad audience, 50.4% of TikTok's and 46.6% of Instagram's. The average user now spends 1 hour 55 minutes a day on TikTok and 1 hour 40 minutes on Instagram. Women are also more likely than men to go online for health information, in every age group.
Indicator | Where | Girls | Boys |
|---|---|---|---|
Problematic social media use, ages 11 to 15, 2022 | 44 countries and regions (WHO Europe) | 13% | 9% |
Say social media hurts their sleep, ages 13 to 17, 2024 | United States (Pew) | 50% | 40% |
Say it hurts their mental health, same survey | United States (Pew) | 25% | 14% |
Say it hurts their confidence, same survey | United States (Pew) | 20% | 10% |
Had nasty or hurtful interactions online, ages 8 to 17, 2025 | United Kingdom (Ofcom) | 34% | 28% |
Problematic social media use, age 13, 2021/22 | Sweden (Folkhälsomyndigheten) | 12% | About 4% |
Sweden shows what a saturated market looks like. Of Swedish internet users aged 8 and up, 83% use social media every day, and more women than men say it reinforces traditional gender roles. In the US, the Surgeon General's advisoryreports that 46% of adolescents aged 13 to 17 say social media makes them feel worse about their body image.
Four design choices, four health risks
Each of these features was built to raise engagement. Each carries a measurable cost, and girls and women pay a larger share of it.
Recommendation feeds amplify harm
Ranking systems optimise for attention, not for health. Independent audits show how fast they escalate. In a 2022 test by the Center for Countering Digital Hate, new TikTok accounts set to age 13 in the US, UK, Canada and Australia were shown eating disorder content within 8 minutes. In Amnesty International's 2025 test in France, three accounts registered as 13-year-olds, two of them girls, had feeds almost entirely filled with mental health content within 15 to 20 minutes.
In 2025 the Molly Rose Foundation ran an account posing as a 15-year-old girl in the UK. After it engaged with posts on depression and self-harm, 97% of recommended Instagram Reels and 96% of TikTok videos were harmful. The industry cannot claim ignorance. A 2019 internal presentation at Facebook (now Meta), reported by the Wall Street Journal, said: "We make body image issues worse for one in three teen girls."
Beauty filters distort the baseline
Filters are a default, not a niche. A Dove survey of 556 US girls found that 80% had used a filter or retouching app by age 13. In a controlled experiment with 144 girls aged 14 to 18, manipulated Instagram selfies lowered body image. The girls spotted the filters but largely missed the body reshaping. In a 2025 experiment with mostly adult women, a slimming filter on their own selfie raised body-dysmorphic thoughts and widened the gap between actual and ideal body.
Like counts turn approval into a score
Teenagers who saw their own photos with many likes showed stronger activity in the nucleus accumbens, a reward centre of the brain. In experiments with more than 600 adolescents, few likes made them feel more rejected, and bullied teens were the most vulnerable. Among 619 adolescents, online comparison and feedback-seeking predicted depressive symptoms, particularly among girls. Instagram made hiding like counts an opt-in setting in 2021. The score stays on unless the user turns it off.
Endless feeds erode sleep
Infinite scroll, autoplay and push notifications remove every natural stopping point. In a UK cohort of 10,904 children aged 14, 43% of girls used social media three hours or more a day, against 22% of boys. At five hours or more, girls' depressive symptom scores were 50% higher than at one to three hours. For boys the rise was 35%. Online harassment, poor sleep, low self-esteem and poor body image carried much of the link.
The pattern holds across borders. A survey of 212,613 adolescents in 40 countries in Europe and North America found that problematic use raised the odds of trouble falling asleep more for girls (2.20) than for boys (1.88).
Health advice, ranked by engagement
The ranking logic that shapes what women see also shapes what they believe about their own bodies. It rewards certainty, emotion and a product to sell. Accuracy is not a ranking signal.
- Contraception. Of 307 TikTok videos on the pill, 54% were negative and 27% came from health professionals. Negative videos averaged 187,876 likes, positive ones 70,488.
- Diagnostic tests. Of 982 influencer posts on medical tests such as AMH fertility tests and full-body MRI, 87% mentioned benefits and 15% mentioned harms. In 68% of cases the account holder had a financial interest.
- PCOS. Physician-made TikTok videos scored 3.6 out of 5 on a modified quality scale. Videos by non-professionals scored 1.0.
- Menopause. Of the top 1,000 Instagram posts on menopause, 66% promoted branded supplements.
The effects reach the clinic. Among women presenting for abortion at BPAS clinics in England, reported use of fertility-awareness methods rose from 0.4% in 2018 to 2.5% in 2023. Use of hormonal methods fell from 18.8% to 11.3%. The authors call the social media link anecdotal. No health system should wait for proof before it answers that content.
The upside is real and must be protected. Platforms gave women a voice on conditions that medicine dismissed for decades. In 34,715 Reddit posts on endometriosis, 49% asked for other patients' experiences. Patients made 83% of the top 100 endometriosis videos on TikTok, and a quarter of those videos highlighted distrust of health care. That distrust is a verdict on health care as much as on platforms.
What the evidence proves, and what it does not
Leadership on this issue requires precision, not alarm. Most of the evidence is correlational. In a 2024 review in Nature, Candice Odgers found "a mix of no, small and mixed associations". She also noted that young people with existing mental health problems may simply use the platforms more. Across 355,358 participants in three US and UK datasets, Orben and Przybylski found that technology use explains at most 0.4% of the variation in adolescent wellbeing. Jonathan Haidt argues the opposite in The Anxious Generation, where he holds that girls took the hardest hit.
Two points are not in dispute. Sensitivity differs by sex and age: in UK data on about 17,400 people aged 10 to 21, heavier use predicted lower life satisfaction a year later at ages 11 to 13 for girls, and at 14 to 15 for boys. And the evidence on adult women is thin. That is a research gap, not proof of safety.
The case for redesign does not depend on settling this debate. A system that serves eating disorder content to a 13-year-old within eight minutes is defective by any product safety standard. No other consumer industry could ship it.
Where regulation stands
Regulation is moving from age limits to design. The most advanced rules now target the feed, the scroll and the notification, not only the user's age.
Jurisdiction | Measure | Status, November 2026 |
|---|---|---|
European Union | DSA guidelines on minors: streaks, read receipts, autoplay and push notifications off by default | Published 14 July 2025 |
European Union | DSA preliminary findings on addictive design against TikTok and Instagram and Facebook | 6 February and 10 July 2026; not final; fines up to 6% of global turnover |
European Union | KIDS Act: no access under 13, own account from 15, burden of proof on platforms | Commission proposal, 17 September 2026 |
United Kingdom | Online Safety Act children's codes: feeds must not push suicide, self-harm or eating disorder content to children | In force since 25 July 2025; fines up to 10% of global revenue |
United States | Meta settlement with 29 states: two-hour daily default for teens, overnight block, likes hidden by default, chronological feed option | Approved 26 August 2026; about $12.2 billion over 10 years; no admission of liability |
United States | Los Angeles jury verdict on infinite feeds, autoplay and notifications | Meta and YouTube found liable, 25 March 2026; $6 million in damages |
New York | SAFE for Kids Act: no algorithmic feed and no night-time notifications for under-18s without parental consent | Takes effect 25 January 2027 |
Brazil | ECA Digital: bans features built to hold children's attention, and ad profiling of minors | In force since 17 March 2026 |
Australia | In force since 10 December 2025; about 4.7 million accounts restricted | |
Malaysia | Minimum age of 16 with ID-based age checks | Enforced since 1 June 2026 |
France | Minimum age of 15 | Struck down by the Conseil constitutionnel on 14 August 2026; new version submitted September 2026 |
Denmark | Minimum age of 15 on large platforms | Draft bill; planned from 1 July 2027 |
Sweden | SOU 2026:35: no logged-in access before the year a child turns 15 | Inquiry proposal, June 2026 |
Almost every measure protects children. Almost none protects adult women, who scroll the same feeds. The EU case against Meta is the exception that matters: it names the risk to vulnerable adults.
Our position
We do not accept damage to women's health as the price of a free platform. Our demands are specific, and each one is already technically possible.
To platforms
- Offer every user, adults included, a non-personalised chronological feed, and respect that choice.
- Label body-altering filters, and keep them off by default.
- Hide like counts by default for all users.
- Label monetised health claims and disclose the commercial interest behind them.
- Make quiet hours at night the default setting.
Meta has already accepted most of these for US teenagers under the August 2026 settlement. What protects a 15-year-old in the United States should also protect a 35-year-old woman in Lisbon, Lagos or Gothenburg.
To regulators
- Extend design duties from minors to all users. The EU case against Meta already points this way.
- Require independent audits of recommender systems, with results reported by sex and age, and fund research on adult women.
To health systems, including our own
- Treat social media exposure as a health determinant. Ask about it, as we ask about sleep and alcohol.
- Close the gap the feed exploits. When women wait months for answers on endometriosis or menopause, the algorithm answers first.
The platforms chose these defaults. They can choose different ones, and the law increasingly says they must. Health care must hold itself to the same standard: when outcomes are bad, fix the design.
