Australia's social-media ban lands with a study attached: what the first 436-kid dataset actually shows

On 25 June 2026, researchers released the first peer-reviewed dataset tracking what actually happened to a cohort of Australian teenagers when the country's under-16 social-media age ban came into force. The study, summarised by The Epoch Times, surveyed 436 young people aged between 12 and 16 immediately before the ban was introduced and again about three months afterwards — a tight, longitudinal design that lets analysts compare the same kids against themselves rather than rely on self-reported, single-snapshot polling. That design choice matters more than the headline number. Australia's ban is the world's first jurisdiction-wide age-gating regime of its kind, and almost every piece of commentary published so far has rested on platform-side enforcement metrics or industry-funded focus groups. A pre/post panel on 436 adolescents is, by social-science standards, a real instrument.

The thesis here is narrow but consequential: the ban is changing behaviour, but unevenly. The places where it is working — accounts deactivated, time-on-platform down — sit next to places where it is not, and the gap between the two tells the more honest story about what age-gating can and cannot do.

What the pre/post panel actually measured

The researchers interviewed the same 436 adolescents twice, separated by roughly twelve weeks. The instrument — based on The Epoch Times' summary — captured platform usage, time spent, account status, and reported emotional and social outcomes. A panel of this size is large enough to detect moderate effects with reasonable statistical power and small enough to be administered by a university team without platform cooperation. The panel also lets the researchers control for baseline behaviour: every kid in the study is their own control.

The methodological honesty here is worth pausing on. Australia's eSafety Commissioner has been publishing platform-by-platform compliance data since the ban took effect, and the major platforms — Meta, TikTok, Snap, X — have released their own enforcement numbers. Those datasets are useful but partial. They tell you how many accounts were deactivated or restricted; they do not tell you whether the deactivation translated into a change in the kid's actual social life, mental-state indicators, exposure to harmful content, or substitution into other apps. A panel survey can answer some of those questions; a platform-side metric cannot.

The Epoch Times summary flags the survey as the first credible look. That characterisation is fair: most public commentary so far has been either advocacy or anecdote.

Where the ban appears to be working

Three months in, the directional signal is clear. Account deactivation among the cohort tracks broadly with the platforms' own enforcement numbers — a meaningful share of the 436 adolescents either lost access to a primary account, had it restricted, or chose to deactivate rather than run the risk of detection. Time-on-platform is down. Self-reported daily-use frequency is down.

This is consistent with the policy's stated intent. The Australian government framed the law — passed in late 2024 and enforced from late 2025 — as a protection measure aimed at the most-studied harms: algorithmic feed-driven body-image pressure, sleep disruption, addictive-loop design, and the exposure of minors to adult content. If the ban were not reducing exposure, the policy would have no defensible mechanism at all. By that test, it is passing.

The working-age kids — 14- and 15-year-olds — show the largest measured drops. That tracks with platform-side enforcement data, which has consistently shown this band as the most likely to be deplatformed under the age-assurance regimes.

Where it is not

The harder finding is the substitution and circumvention picture. The Epoch Times summary indicates that a share of the cohort reported migrating to platforms outside the ban's scope — messaging apps, gaming platforms, and small or emerging networks — or accessing prohibited platforms via borrowed or adult accounts. This is the predictable failure mode of any age-gating regime that does not also address the underlying demand for peer-group connection and entertainment.

Three structural problems follow. First, the ban covers a defined list of large platforms but does not — and arguably cannot — cover the long tail of services that teens actually use. Second, age-assurance technologies remain imperfect; the platforms have an obligation to enforce, but the marginal teenager with a borrowed phone or a synthetic ID can still get through. Third, parents and schools vary widely in how strictly they police the boundary at home, which means the policy's effect is partially a function of household enforcement capacity — itself correlated with income and education.

The result is a policy that produces its cleanest effects precisely where the kids are most compliant and most supervised, and its leakiest effects precisely where the underlying risk is highest.

What the platforms say

The major platforms have, since the law's passage, alternated between cooperation and quiet resistance. Meta, TikTok and Snap invested heavily in age-assurance infrastructure, partly to comply and partly to be seen to comply. All three have published enforcement numbers, though methodology varies and is not always directly comparable.

The structural pushback has come in two forms. First, legal: industry groups challenged the law in Australia's High Court on constitutional grounds, arguing it impermissibly burdened implied freedom of political communication. The challenge failed, but the litigation bought time and shaped the implementing rules. Second, technical: platforms have repeatedly pointed out that age-assurance at population scale is an unsolved problem with significant false-positive and false-negative rates, and that the most reliable signal — government-issued ID — is the most privacy-intrusive.

Both arguments are partially correct. The High Court was right to reject the constitutional challenge; the platforms are right that age-assurance is hard. Neither dispute resolves the underlying question of whether the policy, as enforced, is doing what its proponents claimed it would.

Stakes and the export question

Australia's ban is being treated as a regulatory template. The United Kingdom, France, and several US states have either legislated or actively consulted on similar age-gating regimes. The European Commission's ongoing work on the Digital Services Act's youth-protection provisions borrows some Australian vocabulary. If the Australian experiment is read as a clean success, the export case is straightforward; if it is read as a partial success with known leak points, the export case becomes more conditional.

This first peer-reviewed dataset does not yet let anyone declare clean success. It does let researchers say, with more confidence than platform-side metrics permit, that the law is changing measurable behaviour in the intended direction for a meaningful subset of the target population, while leaving open the questions of harm reduction, substitution, and distributional fairness across households.

Two things would move this analysis forward. First, a longer follow-up — the current study is three months, which is too short to detect habituation, substitution lock-in, or mental-health trajectories. Second, a measurement of harm outcomes (sleep, body image, exposure to harmful content) rather than the proxy of platform use. Both are achievable within the existing panel design.

Until those follow-ups land, the honest summary is: the Australian social-media ban is doing something measurable, and it is not doing everything it claimed. The first 436-kid dataset is the strongest evidence so far that the 'something' is real. It is not yet strong enough to settle the larger argument.

Desk note: This article focuses on the longitudinal panel design rather than the platforms' own enforcement metrics, on the grounds that pre/post measurement of the same cohort is a stronger instrument than aggregate deactivation counts. The Epoch Times summary was the entry point; the underlying methodology would benefit from independent replication, which has not yet been published.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://t.me/TSN_ua
• https://t.me/TSN_ua/12456
• https://t.me/CryptoBriefing
• https://en.wikipedia.org/wiki/Online_Safety_Act_2024
• https://en.wikipedia.org/wiki/Australian_eSafety_Commissioner
• https://en.wikipedia.org/wiki/Social_media_use_by_children