On December 10, 2025, Australia switched on the first national law of its kind: platforms from TikTok and Instagram to Snapchat, YouTube, and Reddit now have to keep under-16s from holding accounts, and by mid-December the country's online-safety regulator reported that roughly 4.7 million such accounts had been blocked or deactivated [source: eSafety Commissioner, 2025]. In the same season, classrooms on nearly every continent were going phone-free. UNESCO counts a fast-rising share of school systems with some form of ban; Florida moved from a first-in-the-nation classroom rule to an all-day, "bell-to-bell" ban for younger grades; England, the Netherlands, and Norway tightened their own policies [source: UNESCO, 2023; source: KFF, 2025].
It looks like one movement. It is really two — and telling them apart is the key to reading the evidence honestly. One targets the phone during the school day. The other targets a minor's ability to sign up for social media at all, anywhere. They share a mood but not a mechanism, and the research behind them is at very different stages.
Two different policies people keep merging
The first policy is the school phone ban: a rule about where and when a device may be used during school hours. It ranges from "away during lessons" to full bell-to-bell prohibition from arrival to dismissal. Its aim is usually framed around attention, classroom behavior, and, increasingly, student wellbeing.
The second is the social-media minimum-age law: a rule about who may hold an account on a platform, regardless of school. Australia's Online Safety Amendment (Social Media Minimum Age) Act, passed on November 28, 2024, sets that floor at 16 and puts the duty on platforms — not on parents or children — to take "reasonable steps" to enforce it, with penalties reaching tens of millions of Australian dollars [source: Australian Government, 2024; source: eSafety Commissioner, 2025].
These are not two names for the same thing. A student can be fully phone-free from 8 a.m. to 3 p.m. and still spend four hours on social media at home; a 15-year-old blocked from opening an account can still borrow a friend's screen at lunch. Conflating them makes the evidence look stronger than it is, because a win claimed for one is quietly credited to the other. Throughout what follows, it helps to keep asking: is this about the device in class, or the account for a minor?
The school-ban wave, by the numbers
The classroom trend is broad and genuinely bipartisan, which is part of why it feels inevitable. UNESCO's 2023 Global Education Monitoring Report urged that technology belong in classrooms "only when it clearly supports learning," noting that in many systems the mere presence of a phone was linked to distraction; at the time fewer than one in four education systems banned smartphones in schools, a share that has climbed sharply since [source: UNESCO, 2023].
In the United States, Florida became the first state to bar phone use during instruction in 2023, then expanded in 2025 to a bell-to-bell ban covering elementary and middle grades; by 2025 more than half of U.S. states had some cellphone law or policy on the books, red and blue alike [source: KFF, 2025]. England's Department for Education issued non-statutory guidance in February 2024 recommending schools prohibit phone use across the whole school day, later sharpening its language toward a phone-free default [source: UK Department for Education, 2024]. The Netherlands introduced a national classroom ban from January 2024 [source: Government of the Netherlands, 2024]. Norway — which a decade ago was handing tablets to first-graders — reversed course, and the great majority of its primary schools now restrict phones.
Politically, then, the consensus is strong. The harder question is what it buys.
What the evidence actually shows
Here is where honesty matters most, because the political certainty runs well ahead of the causal proof. The most-cited recent study on school policy specifically is the SMART Schools research led by the University of Birmingham, published in The Lancet Regional Health – Europe in February 2025. It surveyed 1,227 students aged 12 to 15 across 30 English secondary schools and compared those with restrictive phone rules against those with looser ones. Its finding was sobering for ban advocates: restrictive school phone policies were not associated with better mental wellbeing, nor with better sleep, more physical activity, higher grades, or less classroom disruption [source: Lancet Regional Health Europe, 2025].
That result has to be read for what it is — and what it is not. It is a cross-sectional, observational study: a snapshot comparing different schools at one moment, which cannot by itself establish cause and effect. It does not prove bans are useless; it shows that, on average, schools with stricter rules did not have measurably happier or higher-achieving students. Lead author Victoria Goodyear's own summary was that school phone policies are not a "silver bullet" for wellbeing or behavior.
Now hold that against a study built to get closer to causation. In Norway, economist Sara Abrahamsson used the fact that hundreds of middle schools introduced smartphone bans in different years — a quasi-experimental, event-study design — and linked the timing to national health, grade, and family records. There she found real effects: after a ban, girls' visits to specialists for psychological symptoms fell substantially, bullying declined for both sexes, and girls' grades and likelihood of entering an academic high-school track rose, with the largest gains among girls from lower-income homes [source: Abrahamsson, 2024].
Two rigorous studies, two different answers — and the disagreement is instructive rather than embarrassing. The Norwegian design, tracking the same schools before and after a change, is better at isolating cause than the English snapshot. But it studied full bans in a specific national system, and even it found benefits concentrated in particular groups. A 2024 scoping review spanning 22 studies across 12 countries reached the only conclusion the whole body of work supports: the evidence linking bans to academic and mental-health outcomes is mixed and thin, not strong enough to promise the results politicians often assume [source: Campbell et al., 2024]. The takeaway is not "bans don't work" or "bans work." It is that the design of a study largely determines its answer, and that correlations between a school's rules and its students' outcomes should not be read as proof that the rule caused the outcome.
A ban at school is not a ban on screens
There is a second, quieter finding in the SMART Schools data that matters just as much, because it punctures the most intuitive assumption behind the whole project. Restricting phones at school did not reduce students' overall phone or social-media use across the day [source: Lancet Regional Health Europe, 2025]. The device goes away during lessons; the hours reappear after the bell.
Follow-up qualitative work from the same team described this as compensatory use: some students, cut off during the day, made up the time afterward, sometimes at the expense of sleep or activity. The mechanism is common sense once stated. A school ban governs roughly six hours of a teenager's weekday, on school days only. It does nothing about the bus ride, the bedroom, the weekend, or the summer. If the worry is total screen time — or the content and comparison that come with social media — then a policy bounded by the school gate can only ever touch a slice of it. That is not an argument against classroom bans, which may still be worth it for focus and calm during lessons. It is an argument against expecting them to fix a problem that mostly lives outside school hours.
The bigger claim: is social media causing a mental-health crisis?
Underneath both policies sits a louder claim: that smartphones and social media have caused a collapse in teen mental health. This is the argument of psychologist Jonathan Haidt, whose 2024 book The Anxious Generation assembles a large body of experimental and quasi-experimental studies and lays out mechanisms — lost sleep, social comparison, displaced play — by which phones could be driving the rise in youth anxiety and depression [source: Haidt, 2024].
It is a serious case, and it is also contested by serious people. Reviewing the book in Nature, psychologist Candice Odgers argued that the great bulk of the evidence is correlational, and that the wider literature shows "a mix of no, small, and mixed associations" rather than a clean causal signal; she cautioned against treating the claim that platforms are rewiring children's brains as settled [source: Nature, 2024]. Both scholars are looking at overlapping evidence and weighing it differently — the classic problem of a correlation (teen distress and phone use rose together) that may or may not be a cause.
This debate is not a sideshow to the policy question; it is the foundation being invoked to justify sweeping rules. A reader can hold two things at once without contradiction: that the rise in adolescent distress is real and worth acting on, and that the specific claim "social media caused it" remains genuinely disputed among the researchers closest to the data. Policy is being made in that zone of uncertainty — which is a reason for humility about outcomes, not a reason to do nothing.
The minimum-age experiment
That uncertainty is exactly what makes Australia's law so consequential. Rather than manage the device at school, it tries to keep minors off the platforms entirely until 16, shifting the burden onto companies to verify age. Enforcement runs through "reasonable steps" and age-assurance technology rather than parental consent, which cannot override the floor [source: Australian Government, 2024]. Within days of the December 10, 2025 start, the regulator reported millions of under-16 accounts removed — but also acknowledged the obvious pressure points: teenagers testing workarounds, and age-estimation tools that can be fooled [source: eSafety Commissioner, 2025]. France and others have floated similar under-15 or under-16 thresholds, watching Australia as the test case.
The crucial, easily missed point is that this is a genuine experiment whose result is not yet in. Whether raising the minimum age actually improves young people's mental health at a population level is, at this stage, unproven — Australia is the first natural experiment large enough to find out, and its data will take years. That is a different evidentiary situation from school bans, which at least have a decade of mixed studies behind them. Treating the minimum-age law as a solved solution repeats the same leap the school-ban debate keeps making: mistaking a plausible, popular policy for a proven one. Its promise is real; so is the possibility that determined teenagers route around it, or that the harms migrate to platforms and spaces the law does not cover.
What to watch next
The honest summary is that the world has reached a strong political consensus on top of a mixed and incomplete evidence base — and that the two policies now traveling together will need to be judged separately. For school bans, watch whether the better-designed studies (tracking the same schools before and after, as Norway's did) accumulate, and whether schools pair the device rule with the things that plausibly matter more: what happens at home, digital-literacy teaching, and the long hours outside the gate. For the minimum-age laws, watch Australia's outcome data as it lands, and watch the enforcement question — because a rule that platforms cannot reliably enforce, or that teenagers reliably evade, will not deliver whatever benefit it promises.
None of this is an argument for inaction. Attention and calm in a classroom are worth something on their own, and a society is entitled to decide how young is too young for an algorithmic feed. But the case is strongest when it is made in the currency the evidence actually supports — plausible, partial, worth trying — rather than the currency of certainty. The distinction between a phone in class and an account for a minor, and between a correlation and a cause, is not academic hair-splitting. It is the difference between policy that can learn from its results and policy that merely assumes them.