After a brief moment of excitement upon reading the headlines, a more measured review of the settlement between Meta and 51 US state attorney-generals’ offices (47 states, the District of Columbia, Puerto Rico, American Samoa and the Northern Mariana Islands) leaves us with the bittersweet feeling that the only ones who stand to gain from this case will be Meta’s lawyers and their hefty fees.
The settlement, reached whilst Adam Mosseri, head of Instagram, was still giving evidence, has been finalised for a guaranteed minimum amount of between 12,100 and 12,700 million dollars, payable in ten annual instalments, which could rise to around 17,100 million (18,000 million according to the company’s own accounts) only if TikTok and YouTube adopt equivalent measures and pay out equal amounts. The money will be distributed unevenly amongst the signatory states and will be allocated, as each state decides, to youth mental health programmes, crisis intervention services, out-of-school and summer activities, digital literacy and bans on mobile phones in schools.
The figures
If we look at Meta’s accounts, 12,100 million is equivalent to 18 per cent of the net profit for 2025 (60,458 million). Bearing in mind that it has deferred the payment for 10 years – factoring in the effects of inflation and the advantage of keeping the money in its coffers during those years – the guaranteed annual amount of 1,270 million represents 2.1 per cent of annual net profit, 2.75 per cent of free cash flow and less than 1 per cent of the planned investment in artificial intelligence infrastructure for 2026. In other words, seven and a half days’ worth of Meta’s annual profits.
The measures
This sum (which is not compensation or indemnity) is accompanied by a host of measures to keep minors away from Meta: a daily limit of two hours, combined across Instagram and Facebook, which only a parent can lift, with breaks at 15, 60 and 90 minutes; a night-time block between midnight and 6.00 am; notification silencing from 10.00 pm to 7.00 am and during school hours, from 8.00 am to 3.00 pm, between 15 August and 15 June; an obligation to respond to 90 per cent of reports of harmful content within six hours; a ban on displaying reaction counts and on offering image filters that simulate cosmetic procedures; the option to receive a non-personalised feed, without algorithmic recommendations; robust age verification measures to identify under-18s and ban under-13s; enhanced parental supervision tools (who knows what that will actually turn out to be); an independent audit with broad access to information, including regular reports and the right to refer concerns to the Attorney General; and a court order prohibiting Meta from making false or misleading claims about its safety features.
Most of the obligations are for a period of ten years, but the two-hour limit, night mode and the annual audit are only subject to an initial five-year commitment, which is particularly striking
Most of these obligations are only valid for ten years, but the two-hour limit, night mode and the annual audit are only subject to an initial five-year commitment, which is particularly striking: Meta must consider that these restrictions are, obviously, the ones that cause it the most harm and, therefore, the ones it needs to get rid of as soon as possible. It has not committed – and this is important – to refraining from taking measures that would circumvent these restrictions and allow it to continue exploiting children’s data.
The agreement was reached before Zuckerberg took the stand, but after the testimony of Arturo Béjar, a former engineer in the company’s wellbeing teams. Béjar stated that the safety features were designed not to work, because as they are not enabled by default, the vast majority of users never activate them. He described autoplay, ‘like’ counters and infinite scrolling as unsafe for teenagers. The ‘Take a Break’ tool, for example – presented by Mosseri, head of Instagram, as a success (with over 90 per cent of users reportedly having used it) – is in fact a resounding failure, used by only 1.8 per cent of teenagers. On the witness stand, Mosseri admitted that the tool did not work by a long shot and that the company was well aware of this shortfall. Another cosmetic measure – like many implemented by Meta – designed to fail.
There is a suspicion that Meta has no plans to implement measures that would allow it to win back teenagers in the short term, thereby circumventing the provisions agreed in this settlement. There is already a history of Meta accepting a settlement only to water it down
The brief enthusiasm I referred to at the start is shared only by the attorneys general who have reached the agreement, and is far from unanimous amongst child advocacy groups, who believe that the measures are unlikely to protect children and teenagers given that they can be overridden by parents, will only be audited for five years and are, to a large extent, conditional on the behaviour of the companies concerned. This analysis undoubtedly reveals a mistrust of parents who have proved incapable of setting limits for their children. These are parents whom nobody wants to single out, except Meta, which, through these measures, shifts the responsibility onto them: if children and teenagers remain hooked, it will be their fault.
Background: the sanctions against Facebook
There is a suspicion that Meta has no plans to implement measures that would allow it to win back teenagers in the short term, whilst circumventing those agreed in this deal. There is already a history of Meta agreeing to a settlement and then watering it down. In July 2019, the Federal Trade Commission (FTC) imposed a $5 billion fine on Facebook for breaching the 2012 consent decree, as part of the investigation launched following the Cambridge Analytica scandal. The fine also mandated the creation of an independent privacy committee within the board of directors, whose members could only be removed by a supermajority vote, in an attempt to put an end to Zuckerberg’s sole authority over privacy decisions. Furthermore, it appointed compliance officers who could not be removed by the chief executive, mandated quarterly and annual certifications from Zuckerberg himself (the falsification of which would expose him to personal liability), appointed an independent external assessor (a role undertaken by Protiviti) and established a comprehensive privacy programme with prior review of every new product. It is set to last for twenty years, until 2039.
In 2023, the Federal Trade Commission launched proceedings to prohibit Meta from monetising children’s data; as of today, these proceedings are on hold
Seven years on, the outcome is clear: in May 2023, the FTC itself initiated proceedings to amend the order and prohibit Meta from monetising children’s data, alleging repeated breaches and misleading parents regarding control over their children’s communications. Meta responded by challenging the constitutionality of the Commission itself, following the line taken by the US Supreme Court, which has been dismantling all government agencies (except the Federal Reserve) since Trump was re-elected. To no one’s surprise, the proceedings remain at a standstill as of today. Meanwhile, advertising revenue rose from $69,655 million in 2019 to around $196,175 million in 2025. The data-driven business model was not affected in the slightest.
During the FTC proceedings, two measures were put on the table that would have radically altered the outcome – measures that prompted Zuckerberg to quickly acquiesce to the sanction. The first was Mark Zuckerberg’s personal liability, and the second, the break-up of Meta. Mark owns around 99.8 per cent of the Class B shares, which entitle him to ten votes per share, giving him approximately 61 per cent of the voting power with around 14 per cent of the share capital. This means that no corporate decision is taken against his wishes. The FTC considered taking legal action against him, and there are strong indications that the company paid more than its own lawyers deemed reasonable precisely to prevent its founder from being questioned.
As for the second point, the FTC attempted this on several occasions from 2020 onwards before Trump, with the backing of Zuckerberg and other tech oligarchs, effectively dismantled the federal administration. The request for the divestiture of Instagram and WhatsApp was concluded in November 2025 with a ruling that found Meta did not hold monopolistic power, thereby rendering the request void. The reader may well imagine the current status of the appeal lodged by the FTC.
The Good Samaritan clause
There remains a third measure that the FTC did not put on the table because it exceeds its remit: the repeal or reform of the Good Samaritan clause, Section 230 of the Communications Decency Act of 1996, which prevents the platform from being treated as a publisher of content uploaded by third parties and, therefore, from being held liable for it. It has its origins in a 1959 Supreme Court ruling that found in favour of Smith, a San Francisco bookseller, against a fine imposed by the San Francisco City Council for selling Lolita. He argued that he could not read everything he sold and that holding the supplier liable would have “the collateral effect of inhibiting freedom of expression, making the individual more reluctant to exercise it”.
This protection was extended to a growing internet in which the platforms hosting websites lacked the capacity to review their content, and has brought us to a time where platforms moderate content and carry out algorithmic editing of the timeline by sorting, prioritising and selecting feeds, using artificial intelligence systems capable of reading and classifying everything that is published. The option of removing the liability exemption has already been explored in a 2024 case against TikTok. The judgement in that case concluded that the recommendation algorithm constitutes the platform’s own expression and is therefore not protected by immunity. The Supreme Court, however, declined to rule on this matter in Gonzalez v Google in 2023.
Meta presents the settlement not as a defeat or an admission of wrongdoing, but as the new industry standard
That said, Meta denies having acted improperly. The settlement expressly states that the company accepts no liability towards the claimants. Denying liability shields them – or so they believe – from lawsuits or sanctions outside the United States.
They have had the nerve, if I may say so, to publicise the agreement via an open letter entitled “An Open Letter to TikTok and YouTube to Join Us in Supporting Teens”, in which they present the agreement not as a defeat or an admission of malpractice, but as the new industry standard. Zuckerberg once again presents himself as a visionary, a leader who is showing the way for the competition.
Meta has decided to shift the blame onto the entire sector, making part of the agreement conditional on others joining in and announcing a full-page advertising campaign across various media outlets to put pressure on competitors. Let’s acknowledge that the lawyers who have negotiated this agreement have earned every last dollar they are set to receive.
