On 10 December 2025, Australia became the first country to enforce a minimum age for social media accounts. Platforms including Instagram, TikTok, Snapchat, YouTube, X and Reddit are now required to take reasonable steps to stop Australians under 16 from holding accounts, with penalties of up to A$49.5 million. Other governments are watching closely.
The policy is popular with parents, and the worry behind it is real. But health professionals should be honest about something awkward. The evidence that social media, as a category, is the main driver of the adolescent mental health crisis is much weaker than the political momentum suggests. And a policy aimed at the wrong target can crowd out the ones that would help.
It is worth being precise about what the law does. It does not ban teenagers from looking at content. It places the burden on platforms to take reasonable, privacy-preserving steps to identify under-16 account holders, including on existing accounts, and to remove or deactivate them. That makes it as much an experiment in age verification as in mental health. Whether it changes how young people feel, rather than which apps they use, is an open question nobody can yet answer.
What the data do show
There is solid evidence that some adolescents have a problem. The World Health Organization’s European office, drawing on the Health Behaviour in School-aged Children study of almost 280,000 young people across 44 countries and regions in Europe, central Asia and Canada, reported that problematic social media use rose from 7 per cent in 2018 to 11 per cent in 2022. Girls reported higher levels than boys, 13 per cent against 9 per cent.
“Problematic” here does not mean heavy use. It means addiction-like patterns: being unable to control use, feeling withdrawal, neglecting other activities and suffering negative consequences in daily life. That is a meaningful and growing minority, and it deserves a clinical and public health response.
What the data do not show
What is much less clear is that time on social media, for the average teenager, is a major cause of poor mental health.
In 2019, Amy Orben and Andrew Przybylski of the University of Oxford published an analysis in Nature Human Behaviour using three large datasets. Rather than running one analysis, they ran a very large number of the defensible ways the data could be analysed. Across them, digital technology use explained at most 0.4 per cent of the variation in adolescent well-being. In the same data, the association was comparable in size to that of eating potatoes. Being bullied and using marijuana had far larger negative associations.
Their method also revealed something uncomfortable about the whole field: depending on analytic choices, researchers could produce positive, negative or null associations from the same data. A literature built on such flexible data will produce alarming headlines and reassuring ones in roughly equal measure.
None of this proves social media is harmless. Averages hide the vulnerable minority, and newer platforms and designs have changed since those datasets were collected. But it should make us cautious about treating a ban as the central answer to a problem with many causes.
A small average can still hide real harm
The fairest defence of the ban camp starts exactly here, and it deserves to be taken seriously. A tiny average association is compatible with two very different worlds. In one, social media nudges everyone’s mood down a hair. In the other, it does nothing to most teenagers and real damage to a minority. The second world would produce a small average too.
Run the arithmetic. Suppose, purely for illustration, that 90 per cent of teenagers are unaffected and 10 per cent experience a meaningful decline in well-being. Averaged over the whole population, that effect shrinks by a factor of ten and can look trivial in a large survey. Now put it next to the WHO Europe figure: 11 per cent of adolescents with problematic use. The overlap between that group and the “harmed minority” is unknown, but the numbers are the same order of magnitude, which is precisely why averages alone cannot settle the argument.
This is where the counterargument partly holds. Critics who cite the Orben and Przybylski result as proof of harmlessness are overreaching. The small average does not clear the platforms. What it does undermine is the policy logic of a blanket ban. If harm is concentrated, a measure that removes access for every under-16 imposes costs on the majority to reach a minority it cannot identify, and it does nothing for the 16- and 17-year-olds who fall on the other side of the line. Concentrated harm argues for finding the harmed, not for switching off the category.
There is a causality problem too. Teenagers who are already anxious, lonely or depressed may use social media differently and more heavily. Cross-sectional surveys cannot tell which came first. That is not a technicality. If distress drives problematic use as much as the reverse, then removing the app treats a symptom and leaves the cause untouched.
The risk of the convenient villain
Adolescent mental health is shaped by poverty, family conflict, bullying, academic pressure, sleep, substance use, loneliness and access to care. Most of those are expensive and politically difficult to fix. A social media ban is cheap for governments, costs platforms more than taxpayers, and is easy to explain.
That is the uncomfortable incentive. A policy can be popular precisely because it shifts attention and responsibility away from harder investments, like school counsellors, child and adolescent mental health services, and waiting lists that in many countries stretch for months.
There is also a trade-off for particular groups. For some isolated young people, including LGBTQ teenagers and those with rare conditions or disabilities, online communities are a source of support that is not available locally. A blanket age limit removes that along with the harms.
What the ban does to incentives
Consider how the law reshapes behaviour for each party. Platforms face large penalties for failing to take reasonable steps, so their rational response is to invest in age assurance and account removal, the thing they are measured on, rather than in redesigning feeds for the teenagers who remain or slip through. A compliance regime rewards proving absence, not reducing harm.
Teenagers, meanwhile, are resourceful. Some will lie about their age, borrow a parent’s account or move to smaller services and messaging apps that sit outside the law’s scope and moderation budgets. Displacement is the classic second-order effect of prohibition: use falls on the monitored channels and reappears where adults see less of it. That makes any evaluation that simply counts under-16 accounts on named platforms almost meaningless.
Parents may relax. That sounds benign, but a household that believes the government has handled the problem is less likely to have the harder conversations about sleep, bullying or what a child is actually doing on a phone. And governments gain a talking point that can stand in for the unglamorous spending that would help most.
None of this means the Australian experiment is worthless. It is genuinely informative, and it may well reduce some harms. The danger lies in treating it as settled policy to be copied before anyone knows what it did.
Better targets than the account itself
If the aim is fewer adolescents harmed by their digital lives, the tools differ by who holds them.
Regulators can target design, not just access. Infinite scroll, algorithmic amplification of harmful content and engagement-maximising notifications at midnight are choices companies make, and rules can require defaults that are safer for minors whatever their age on paper. Design duties also apply to 16- and 17-year-olds, whom an age limit ignores.
Clinicians can screen for problematic use rather than screen time. Asking about loss of control, withdrawal and consequences distinguishes harm from heavy use, and it identifies the minority the averages hide. Paediatricians and school nurses who ask those questions routinely will find more struggling teenagers than any age-verification system.
Families and schools can protect sleep. Night-time use displacing sleep is one of the more plausible pathways to harm, and phones charging outside the bedroom is a low-tech intervention with a plausible mechanism and almost no downside.
Health ministries and funders need to fund the services. A teenager identified as struggling needs somewhere to go, and in many countries waiting lists for child and adolescent mental health care are the real bottleneck. Every screening programme that finds more children without adding capacity simply lengthens the queue.
Researchers, finally, should treat Australia’s policy as the natural experiment it is. That means outcomes defined in advance, including mental health measures rather than account counts, comparison groups, and tracking of where young people’s online time went instead.
The age-verification wager
There is a second experiment running inside the first. To enforce an age limit, platforms must estimate the age of millions of users, including adults, through some combination of self-declaration, behavioural signals, facial age estimation and document checks. Each method trades accuracy against privacy. A system strict enough to keep out determined 14-year-olds will inconvenience or misclassify a meaningful number of adults, and one light enough to avoid that will leak.
This matters for health beyond the teenage question. Once the machinery for proving age online exists, it tends to be reused for other purposes, and the privacy costs fall on everyone. For founders building age-assurance tools, the law creates a market. For health professionals, the question is narrower and harder: whether whatever mental health gain the ban produces is worth that infrastructure, and whether the same political capital could have bought design rules that protect all minors without it.
Technology companies are not neutral observers here either. A ban framed around access lets platforms argue that they have met their duty once the under-16s are gone, and it shifts attention away from product features that also affect older teenagers and adults. Design regulation is the thing they have most to lose from, which is one reason it is worth pursuing.
What to watch
The next 12 to 24 months will produce the first real data. Watch whether evaluations of the Australian law report mental health outcomes or only compliance statistics; whether other governments copy the age limit before those results exist; whether regulators pair access rules with enforceable design duties; and whether any government announcing a ban also announces money for adolescent mental health services. The last one is the best single test of whether a ban is a health policy or a substitute for one.
The takeaway
The adolescent mental health crisis is real, and some young people are clearly harmed by how they use social media. But a ban is not a treatment plan. If governments use age limits to declare the problem addressed, while leaving mental health services underfunded and platform design untouched, they will have picked the most visible villain rather than the most important one. The better path is less satisfying to announce: fix the design, find the harmed minority, fund the services and measure honestly. It will not produce a single dramatic date. It is more likely to produce teenagers who are actually better off.
Would you support an under-16 social media ban where you live, and what evidence would change your mind? Tell us in the comments.
