The Settlement Illusion: The $18 Billion Mirage
When the 29 states suing Meta in a California court started their proceedings there was talk of a billion in damages. So why was $18bn described by Rob Bonta, the California Attorney General who led the case, as “transformational.” Meta called it proof it was “partnering with parents and experts.” Read past the headline, though, and the deal looks considerably less impressive than the number suggests.
Thirty five years ago a group of state attorneys were suing Lloyd’s for a range of breaches of state and federal laws. As Names we were very pleased to have their support and I, then Chairman of the Lloyd’s Names Association, spent some time briefing them and answering their questions. They were very sure of their legal grounds and went off to negotiate with Lloyd’s. In a puff of smoke their determination disappeared and with a few concessions to support the US Names (at the expense of other Names) the threatened litigation was abandoned and the attorneys declared a settlement victory while Lloyd’s breathed a sigh of relief. I was therefore not surprised when this much vaunted litigation against Meta ended after a week with a deal. Attorney-Generals in States are elected officials; they are interested in quick results they can boast about when seeking re-election or moving on to be candidates for governorships. A case like the one against Meta that generates big headlines is a good one for them, and a quick ‘win’ in a settlement preferable to years of appeals and wrangling in courts. So why did this case outcome matter so much?
Start with what was actually agreed that will improve life for children. Teen accounts on Facebook and Instagram will default to a two-hour daily limit, a midnight-to-6am block, and reduced notifications during the school day; only a parent will be able to lift any of these settings. Like-counts will be hidden by default. Meta will fund an independent auditor to check its own compliance. These are genuine, welcome changes — the kind child safety campaigners have been asking for since Frances Haugen’s testimony to Congress in 2021. But it is worth noticing what is missing: nothing in the settlement touches the recommendation algorithm itself, the engine that decides what a child sees and how long they stay looking at it. Arturo Béjar, the former Meta safety engineer whose testimony helped build the states’ case, put it bluntly after the deal was announced: the agreement risks becoming little more than “safety theatre.” His analogy was pointed — restricting a child to two hours a day, he said, is like being allowed only two hours of cigarettes a day. It doesn’t make the product any safer. It just limits how long you’re exposed to it.
Then there is the money itself. Meta will pay out over ten years, in instalments of roughly $1.17 billion a year — a sum reportedly comparable to what the company’s Reality Labs division loses every twenty-four days. Set against Meta’s $60 billion-plus in annual profit and its close to $1.5 trillion market valuation, the settlement represents well under 1% of annual revenue. Analysts covering the stock have been candid about what this means: one described the deal as “manageable” with “minimal impact on long-term profitability.” None of this is a reason to dismiss the changes altogether. But it is a reason to be sceptical of language like “historic” and “transformational,” and to ask what a truly historic penalty would actually need to cost a company of this size to change its behaviour rather than its PR.
Not every state was persuaded. Florida refused to join the settlement and is continuing to pursue Meta through the courts. Its attorney-general, James Uthmeier, was characteristically blunt: trying to “wipe out a decade of harm to the nation’s youth with one month’s cash flow,” he wrote, was “an insult.” New Mexico did not join either — not out of protest, but because it already holds its own judgments against Meta worth nearly $1 billion between a March jury verdict and a further court-ordered payment in August, both of which Meta is now appealing. Two very different reasons for standing apart, but both point to the same conclusion: the $18 billion figure is a negotiated ceiling (and given, some is not payable unless competitors agree, and it will be paid over a decade so in discounted terms is worth more like $8bn), not a natural one, and other routes to a bigger reckoning remain open.
The structure of the payment itself repeats the same pattern. Only 70% of the $18 billion — around $12.7 billion — is guaranteed. The remaining 30%, roughly $5.3 billion, is released only if TikTok and YouTube agree to adopt equivalent restrictions, a domino logic borrowed explicitly from the tobacco settlement, where every major manufacturer signed at once. But tobacco had four companies and one product. Social media has dozens of platforms, endlessly substitutable in a teenager’s attention, and no mechanism in this settlement forces TikTok or YouTube to do anything at all — Meta can only ask, and has done so publicly, calling on rivals to “join right away” – but why should they when their commercial interest lies in continuing to provide what Meta won’t be? The other concessions read the same way on close inspection: personalised, algorithm-driven feeds can be turned off for a teen, but again, only if a parent actively requests it — the system stays switched on, by default, for every child whose parent doesn’t know the setting exists, or can be persuaded by their children to turn it off.
There are right-wingers in the States who argue that any restrictions on Big Tech are an infringement on the right to free speech. That right has never been an unguarded right. Any responsible society has put curbs on that right to protect vulnerable citizens from abuse (racial, sexual or whatever) and from inflammatory language intended to cause trouble. Children are the most vulnerable group in society, most needing the protection we give them. History will look back with amazement at the ease with which Big Tech was allowed to experiment with our children’s brains – and to build nudge systems to manage all of us.
There is also the practical question of whether time limits of this kind actually work on teenagers. Australia’s experiment is instructive. Having banned social media outright for under-16s, regulators there are now finding that large numbers of teenagers have simply found their way back onto the platforms — by one estimate, roughly eight in ten young teens remain on social media despite the ban, and specific platforms like TikTok have seen usage climb back toward pre-ban levels within months. If an outright legal ban cannot keep Australian teenagers off these apps, it is not obvious that a two-hour timer with a parental override — and an obvious incentive for a restricted teenager to simply migrate to TikTok or Snapchat instead — will do much better.
The comparison several commentators have reached for is the 1998 Master Settlement Agreement between the major tobacco companies and 46 US states. It is an apt parallel in one sense: both cases followed the same arc of internal documents, whistleblowers and years of public denial, followed by a large settlement that changed the industry’s public conduct without ending its core product. But it is worth remembering how long that earlier settlement took to matter. It reshaped tobacco marketing and funded public health campaigns for two decades before smoking rates fell meaningfully, and it did so backed by continuous, independently verified monitoring of the industry’s behaviour. Meta’s settlement includes an auditor, but not the algorithm-level transparency Béjar and others say is actually needed to know whether children are safer, or simply logged off for two fewer hours a day while the underlying system stays exactly as it was.
None of this means the settlement is worthless. But it is not the reckoning the headline number implies, and it should not be mistaken for one. The hard evidence — from Australia’s bypassed ban, from Béjar’s own testimony, from the sheer scale of a $1.5 trillion company’s annual profit — suggests that what actually protects a child from the attention economy still has very little to do with what a court or a regulator can mandate on the platform itself, and a great deal to do with what happens to that child away from the screen altogether.
What really matters about this case is not the settlement – a mirage. It is the discovery that preceded it. For the next decade that discovery will be mined in successor litigation as people, states and other national governments start the process of restricting Big Tech’s exploitation of all of us – particularly our children. There is still a long march ahead for proper regulation. In the meantime parents must struggle alone to control a highly addictive product and help children find alternative passions.
