Meta’s $12b US child safety settlement won’t protect our kids

Every evening, in hundreds of thousands of New Zealand homes, children sit bathed in the glow of screens running apps that were deliberately engineered to keep them scrolling. The algorithms don’t care that it is a school night, or that the child is 13, or 11, or nine. They’re optimised for one thing: attention. Which, in the economics of social media, is revenue.

Last week, in a federal courtroom in California, this business model met a moment of public reckoning. Meta, the company that owns Instagram and Facebook, agreed to pay at least US$12 billion to settle claims brought by 29 American states. It’s the largest accountability event of social media so far. It has been called Big Tech’s Big Tobacco moment. The comparison is apt, and not only in the ways that flatter it.

The trial in California lasted barely two weeks before Meta pulled the plug and settled. The reason became obvious from the evidence. A whistleblower, Arturo Béjar, produced an email he had sent directly to Mark Zuckerberg in 2021. It showed that 51 percent of teenage users reported a harmful or distressing experience in the previous seven days, including bullying, sexual content, and self-harm material. The platform removed content on just 0.02 percent of occasions.

An Instagram design director admitted that data had been deliberately cut from a slide deck before it was shown to senior leadership. The missing data showed that teenagers encountered bullying, hate speech, nudity and violent content at one and a half times the rate of adult users.

Instagram’s head, Adam Mosseri, gave evidence on a Monday. Meta settled the next morning. Its share price barely moved. There was no admission of wrongdoing. The company agreed to impose a two-hour daily limit for teenage users, block overnight access without parental consent, and mute notifications during school hours. An independent auditor and a research foundation were thrown in. For a company worth well over a trillion dollars, it was the cost of doing business.

Here is the uncomfortable truth for New Zealand: we have no equivalent law under which a penalty of that magnitude could be extracted. There is no institutional equivalent of attorneys general with their own enforcement budgets and political incentives to act. New Zealand’s privacy regulator’s toolkit is limited, with a maximum fine of a $10,000. Consumer protection fines max out at figures a company like Meta wouldn’t notice on a quarterly earnings call. We might get some of the practical benefits by accident. Product changes agreed to in California might take place also in New Zealand. But receiving a spillover benefit from someone else’s lawsuit isn’t the same as holding a company accountable. It’s luck, not governance. And luck isn’t a regulatory strategy.

What made California’s proceedings so devastating was not that they revealed something new. It was that they put on the record, under oath, what parents, teachers, researchers and clinicians have been saying for years and proved that the companies knew it too.

The internal documents read into the court record were not the work of outside critics. They were Meta’s own findings, generated by Meta’s own product teams, showing in granular detail that the platforms were harming young people and that company leaders knew it. Infinite scroll, autoplay, beauty filters calibrated to distort self-image, recommendation engines that serve progressively more extreme content are features built to maximise the time a user spends on the app. Time is what advertisers pay for.

New Zealand’s children are subject to the same engagement architecture as their American counterparts.

A parliamentary inquiry into online harms affecting children delivered its final report here in March 2026, making 12 recommendations for legislative action. Two days before the Meta settlement, a new Online Safety Bill was introduced to Parliament proposing a minimum age of 16 for social media accounts. The concern isn’t imported. It’s homegrown, and urgent.

But the most important lesson from California may be what the settlement did not achieve. It didn’t require Meta to abandon personalised recommendations. It didn’t touch the advertising model that funds the entire operation. The revenue engine that drives the harm remains intact.

The parallel with tobacco isn’t just rhetorical, it’s structural. The 1998 Master Settlement Agreement a landmark legal settlement between 46 US states and the major tobacco companies changed how cigarettes were marketed, but it left the cigarette on the shelf. Twenty-eight years later, you can still buy one. Unless regulation addresses the algorithm itself, and not just the conditions under which children access the platform, the engagement loop will keep spinning long after the headlines fade.

If the social media debate feels urgent, the artificial intelligence conversation should feel more so.

This is not speculative. The chief executives of the companies building the most powerful AI systems have said, publicly and repeatedly, that the technology poses serious and immediate risks to society. Bill Gates has issued similar warnings, even stating that Big Tech lies about the risks. When the people constructing these tools are sounding the alarm, it takes a particular kind of complacency to assume the rest of us shouldn’t worry.

Generative AI produces disinformation, manipulated imagery and harmful content at a speed and scale no human moderator can match. Algorithmic systems are quietly shaping who gets a loan, who sees a job advertisement, and what a teenager is shown next, with little transparency about how those decisions are made.

New Zealand’s Online Safety Bill does have some genuine strengths. Its proposed penalty regime, fines of up to 10 percent of a platform’s global revenue, is the kind of sanction that scales to the size of the companies involved. Age verification requirements, public reporting obligations, and the explicit inclusion of AI companions within the bill’s scope all represent serious policy thinking. The problem isn’t in the drafting. It is the politics.

As demonstrated in the United States, voluntary self-regulation doesn’t work.

Litigation, however spectacular the headlines, is not a substitute for legislation, not in New Zealand. Meta settled in California to limit its legal exposure, and to also limit their competitors. New Zealand has never been particularly effective at courtroom confrontation with trillion-dollar companies.

Regulatory alignment, moving with Australia, the UK and the EU, would make compliance cheaper than a fight. But alignment requires action, and that takes political will.

New Zealand’s children are online now, tonight, subject to the same algorithms and the same extraction of their attention and their data. The moment of accountability that 29 American attorneys general forced in California can come in New Zealand only from Parliament. If it doesn’t come soon, the next generation will bear the cost, and no accidental benefit from someone else’s lawsuit will be enough to excuse it.