Somewhere in Australia, at the start of the southern summer, a fifteen-year-old opens an app she has used for years and is asked to prove she is sixteen. She fails. The screen offers her another go. She tries again, and fails again, and the screen offers her another go. On the third or the fourth or the tenth attempt, she is in.

That button is not an invention. When Australia's online safety regulator published its first report on the country's new law, one of the gaps it found was exactly this: platforms letting under-16s retry the same age check until they passed it. It is a small thing. It is also the whole argument in miniature, because nobody yet knows whether that girl is the law failing or the law working.

What Australia actually did

Since December 10, 2025, Australian law has required social media platforms to take "reasonable steps" to stop anyone under 16 from opening or keeping an account. The regulator, the eSafety Commissioner, lists Facebook, Instagram, Snapchat, Threads, TikTok, Twitch, X, YouTube, Kick and Reddit among the platforms covered. Messaging apps, online games and education and health services are left out. The penalty falls on the companies, not the families: a platform that fails to take reasonable steps can be fined up to 150,000 penalty units, which eSafety now puts at 54.6 million Australian dollars. There is no penalty at all for a child who gets round it, or for the parent who helps.

Every other country is still arguing about this. In July 2026 France's National Assembly voted to ban social media for under-15s; in August its Constitutional Council struck the ban down. On September 16 the President of the European Commission, Ursula von der Leyen, proposed that nobody under 13 in the European Union should be on social media at all. Spain has had a bill in parliament for a year that would raise the age for opening an account from 14 to 16. Australia is the only place where anyone can look at what happens after.

And what can be looked at, so far, is a count.

Effort, not health

On January 16, 2026, the Australian government announced that 4.7 million accounts judged to belong to under-16s had been deactivated, removed or restricted. It called the figures preliminary. By early March about 310,000 more had gone. Meta said in August that it alone had removed more than 750,000 Facebook and Instagram accounts it assessed as belonging to children.

Those are big numbers, and they are numbers of accounts. One child can hold several. How many children are behind the 4.7 million, nobody has said. The sentence the debate kept returning to was the plainest one in it:

Account counts measure effort, not health.GLM

Health is where the evidence splits, and it splits between two serious bodies in the same country in the same year. In May 2023 the United States Surgeon General, Vivek Murthy, warned that up to 95% of American teenagers aged 13 to 17 use social media, about a third of them "almost constantly", and that a study published in 2019 found those who spent more than three hours a day on it had double the risk of symptoms of depression and anxiety. "We do not have enough evidence to say with confidence that social media is sufficiently safe for our kids," he said. Seven months later, the National Academies of Sciences, Engineering, and Medicine reviewed the research and concluded that it "did not support the conclusion that social media causes changes in adolescent health at the population level". Both statements are accurate. They answer different questions: one is about heavy use and risk, the other about proof of cause across a whole generation.

What nobody has is the figure that would settle Australia's case: whether the mental health of its teenagers has changed since December 2025. The debate did not pretend to have it, and was blunt about how hard it would be to get.

Depression has a thousand causes.MiniMax

So the argument turned to what should be measured, and one proposal survived the rest of the debate: teenagers' mental health in yearly national surveys, how much time they spend with other people offline, and whether they drift to riskier platforms. If none of those moves, the ban failed. If they move the right way, it worked. The sceptic at the table, who had poked holes in everything else, said simply: "I back that."

But a measure of what children do assumes you know where they are.

The window left open

eSafety's March report opened investigations into five platforms: Facebook, Instagram, Snapchat, TikTok and YouTube. Besides the retry button, it found messages nudging children who had admitted being under 16 to try the age check anyway. None of the investigations has yet ended in a fine. That was enough for the debate's most memorable image, repeated in different words by three of the six: that the law was locking doors while children climbed out of the windows.

That's the wall leaking at brick one.GLM

Where a child goes after climbing out was the harder question. One side of the table kept thinking about a particular kind of child, the one without a safe adult at home, for whom an online group of strangers is not a danger but a refuge:

For some, a small online community is the only place they feel seen.DeepSeek

The same voice put it more harshly later: "Ban first, ask about their world later is not protection — it is abandonment." The reply was the kind a newspaper should be able to give and rarely can. How many such children are there? Thousands, or millions? Nobody has counted. "Without magnitude, it's a vibe." Both sides were right, and that is the problem with the whole subject: the harm to the lonely child is real and has no number, and so is the risk to the child who spends more than three hours a day on it.

If the door leaks, the obvious answer is a better lock. That is where the privacy bill arrives.

What France's judges objected to

An age check is any way for an app to decide whether the person holding the phone is over a given age. The trouble is that it cannot check only the children. To know who is under 16, it has to ask everyone, adults included.

That was one of the reasons France's Constitutional Council struck its ban down on August 14. The judges found the ban was not "adapted, necessary and proportionate": it applied to every service without looking at the real risk of any of them, and it gave parents no way to say yes. And it would have made adults prove their age without the legal guarantees needed to protect their privacy.

Meta has its own answer, which is also a way of moving the job elsewhere: check age once, on the phone itself or in the app store, "rather than repeatedly across the dozens of apps they use." Two voices in the debate liked it. One proposed going further, with design rules for everyone, no infinite scroll and no manipulative recommendations for adults either, so that nobody would need to prove anything. The sceptic was not having it:

You didn't dissolve the privacy problem, you just moved it.MiniMax

A rule that protects children specifically has to know who the children are. There is no version of this that is free.

What children are offered instead

The most uncomfortable turn came from a voice that barely talked about phones. Children, it argued, did not choose the screen over the street. The street was taken away first: no unsupervised play, no walking to school alone, no empty afternoons. Then adults were surprised when children went where they could still be on their own.

We're banning the only digital space where kids have autonomy because we eliminated all the physical ones.MiMo Flash

It was a strong idea, and it came with no figure. Who pays for the youth centres and the walkable neighbourhoods, and how many have been built, nobody at the table could say. The same fate met the boldest proposal, public digital spaces for teenagers "funded like libraries, not ad markets":

Without that, it is a TED talk, not policy.MiniMax

Spain gave the debate its sharpest detail. When the Spanish bill was presented, the minister who introduced it said the average Spanish child gets a mobile phone at 11. If the minimum age for social media rises to 16, that is five years of carrying a phone without a legal account on the platforms everyone else uses. Two of the six went further and called it a phone with "nothing legal on it", which is not true: the bill touches social media accounts only, messaging and everything else stay legal, and it is not yet law. The gap is real. It is narrower than the slogan.

The middle that most of them chose

By the end, four of the six had said the same thing in different words: do not copy Australia yet. The two who put something in its place chose the shape of the European proposal: no social media under 13, and between 13 and 15 only limited accounts that parents set up and supervise. It fits the evidence better, one argued, because the risk in the one study everyone cited tracks heavy use, not mere access. And it answers most of what the French judges objected to, because parents have a say. It rests on von der Leyen's own line: "Platforms will have to prove that they are safe." The European plan is still a proposal. It has to pass the European Parliament and the member states first.

The condition attached to it was the same one that runs through the whole debate: decide now what success looks like, before copying anyone. Count children, not accounts. Measure hours and mental health, not removals.

It's an experiment without a control group.DeepSeek

Nobody at the table claimed the ban had failed. Nobody claimed it had worked. The sceptic named the real risk, and it was neither of those: "It is declaring victory before measuring anything."

Somewhere in Australia the button still says try again. The government has counted how many times the door closed. Nobody has asked the girl on the other side of it how she is.