Eleven Days of Profit

What Meta Knew, Why the Settlement is So Large — and Why Florida Was Right to Walk Away From It

On August 26, eight days into a trial in Oakland, Meta Platforms stopped fighting.

The case had been brought by twenty-nine states, with the claims of California, Colorado, Kentucky and New Jersey tried first as a bellwether before Judge Yvonne Gonzalez Rogers and an eight-member advisory jury. The states alleged that Instagram and Facebook were engineered to capture children and hold them, and that Meta told the public otherwise. The evidence was still going in when the parties filed a proposed consent judgment, and the judge entered it before the day was out. By that afternoon there was a number, and the number was all anyone talked about.

It is worth understanding what that number actually is, what it took to produce it, and — for families in this state — why Florida is not in it.

What is actually being paid

Take the figure from Meta’s own disclosure rather than from the headlines. Meta describes a payment of approximately $18 billion. Roughly 70 percent — about $12.7 billion — goes to participating states over ten years. The remaining 30 percent, about $5.3 billion, is released only if YouTube and TikTok adopt matching limits and each pay a matching amount of their own. If they never do, that money is never paid. The filed multistate youth settlement tops out at approximately $16.68 billion. Some official descriptions use roughly $17.1 billion after including a separate $459 million Cambridge Analytica resolution. Meta, describing its broader agreement with 52 attorneys general, puts the total at approximately $18 billion, and Meta’s own figure is the one used here.

So the guaranteed sum is roughly $12.7 billion across a decade — about $1.27 billion a year.

In 2025 Meta reported net income of $60.46 billion on revenue of $200.97 billion. That is roughly $166 million in profit every day. Which puts the guaranteed annual payment at just under eight days of Meta’s profit. If every contingency triggers and the full amount is paid, it is about eleven days.

Eleven days of profit a year, for ten years, to resolve the claims of nearly every state in the country.

None of that is concealed. Meta published the structure itself, alongside an open letter urging TikTok and YouTube to join. But arithmetic does not travel the way a headline does, and the space between the two is where most of the public understanding of this week now sits.

Why the number is as large as it is

A settlement of this size is not generosity. It is a price set by what was about to happen in that courtroom.

Two months earlier, on June 29, Judge Gonzalez Rogers denied every one of Meta’s summary judgment motions and granted partial summary judgment that Meta had not complied with the notice and parental-consent requirements of the Children’s Online Privacy Protection Act — while leaving for trial whether the statute applied to Meta at all. She wrote that Meta’s denial “strains credulity,” and that Meta “has repeatedly urged that it need not comply. Based on this position, it has not.” She also found that Meta’s own documents supported the states’ theory that its time-limit tools were merely a “public relations stunt.” The states had assembled more than a hundred statements by Meta representatives they contended were deceptive.

Then the trial started, and the documents began going in front of the jury. The states’ opening described a business model of four steps: hook the users, hold them, harvest their data, hide the truth. Jurors were shown a 2016 internal email identifying “teen time spent” as the overall company goal for Instagram, and an internal study titled “Long Term Retention: The Young Ones Are The Best Ones,” which examined ten-to-twelve-year-olds and concluded that the earlier a user starts, the longer they stay and the more revenue they generate.

Arturo Béjar, a former Meta product safety engineer whom the company had tried and failed to keep off the stand, testified that in a study of more than two hundred thousand users, “we found that younger people reported higher rates for almost every issue,” and that Meta’s published safety metrics “create a false impression of safety.” On August 25, Adam Mosseri, the head of Instagram, was confronted with an internal document showing that only 1.8 percent of teenagers ever used the “Take a Break” feature Meta had promoted publicly. He acknowledged the company had not disclosed that figure.

Meta settled the next day.

And it did so knowing what juries had already done with this evidence. In March, a Los Angeles jury in a single young woman’s case found Meta and YouTube negligent and their products defectively designed, apportioning seventy percent of the fault to Meta and awarding six million dollars including punitive damages — the first jury verdict in the wave of social-media product-design litigation. One day earlier, a New Mexico jury found seventy-five thousand violations of that state’s Unfair Practices Act and imposed the maximum penalty. In August, the New Mexico court went further and ordered Meta to fund a public nuisance abatement program. Meta is appealing both.

Meta expects to record roughly $10 billion in legal expense in the third quarter of 2026 alone. That is the context in which eleven days of profit a year began to look like a bargain.

What Meta knew, and what it says about it

The states’ case rested substantially on Meta’s own research, and the fairest way to describe that research is in the company’s own published words rather than a journalist’s.

After the 2021 Wall Street Journal reporting drawn from Frances Haugen’s disclosures, Meta itself released annotated copies of two 2019 internal Instagram decks. Among the material Meta published is a slide stating that the company makes “body image issues worse for one in three teen girls.” Other internal language: “teens blame Instagram for increases in the rate of anxiety and depression” — noted as “unprompted and consistent across all groups.” Another slide: “social comparison is worse on Instagram.” And, from the second deck, a researcher’s summary that teens “often feel ‘addicted’ and know that what they’re seeing is bad for their mental health but feel unable to stop themselves.”

Meta’s response — in a published rebuttal and in its annotations on the decks themselves — deserves to be stated with equal precision, because parts of it are well taken. The company says the “one in three” figure describes one in three of those teen girls who had already reported body image issues — not one in three of all teen girls — and that fewer than a hundred and fifty girls answered those particular questions. It says that on eleven of twelve difficult issues, teenage girls who struggled reported that Instagram made things better rather than worse, with body image the exception. It says the research was designed to surface the worst problems precisely so they could be fixed, which is why the harshest findings appear on the slides. And it rejects the word “toxic” as a headline writer’s, not its own.

That is a real defense. It is also, notably, not a denial that the findings existed.

The states alleged considerably more — that Meta knew millions of children under thirteen were on its platforms and collected their data anyway, that its own researchers were “confident of a causal link” between visible like counts and social comparison, and that internal goals were set around increasing teen time spent while executives publicly denied having such goals. Those are allegations. Meta denied them, and the consent judgment resolves them without any admission of liability. But the judge’s COPPA ruling in June was not an allegation. It was a finding.

What the evidence outside the courtroom actually shows

A lawyer should be careful here, because this is a field where the confident public conversation runs well ahead of the science.

The strongest causal evidence available is a 2022 study in the American Economic Review by Braghieri, Levy and Makarin, which used the staggered rollout of Facebook across American college campuses as a natural experiment. Because Facebook arrived at different schools at different times for reasons unrelated to student mental health, the researchers could compare like to like. By the authors’ own account, campus-wide access to Facebook increased the number of students reporting severe depression by seven percent and reported anxiety disorder by twenty percent, with the mechanism appearing to be unfavorable social comparison. That is a quasi-experimental design, not a randomized trial, and it studied college students on early Facebook rather than adolescents on modern Instagram.

The public health bodies have moved in the same direction with appropriate hedging. The Surgeon General’s 2023 advisory concluded there are ample indicators that social media can pose a profound risk of harm to the mental health of children and adolescents, while stating plainly that the evidence is not yet sufficient to declare it safe or to quantify the harm precisely. The American Psychological Association issued a health advisory the same year focused on developmental vulnerability rather than screen time totals.

And there is serious dissent, which belongs in any honest account. Candice Odgers, Amy Orben and Andrew Przybylski have argued for years that the large correlational studies show associations too small to be clinically meaningful and too tangled to establish direction. Orben’s work makes a narrower and more useful point: that the question is not how many hours but which mechanisms, at which stage of development, for which child.

The honest summary is this. The claim that social media is uniformly harmful to every child is not established. The claim that specific engineered features can harm specific vulnerable adolescents has real support — and it is the claim these lawsuits actually make.

That distinction matters more than it may appear. A design case does not require proving that a product harms everyone. These cases turn on particular features: whether they created an unreasonable or foreseeable risk, whether the law holds the company responsible for that design, and whether the feature caused the plaintiff’s injury.

Why Florida walked, and why that was the strong call

Attorney General James Uthmeier declined this settlement outright. He called the payouts “peanuts” measured against the harm alleged and “a slap on the wrist for a trillion-dollar corp,” and said Florida will see Meta at trial. Florida’s procedural position deserves to be stated exactly, because it is unusual. Florida sued Meta in federal court in Tampa in October 2023 under its consumer-protection statute and the federal children’s-privacy law. The case was swept into the consolidated multidistrict proceeding in California a month later. Florida litigated there for roughly a year before voluntarily dismissing its action without prejudice in December 2024. As of this writing, no action brought by the Florida Attorney General against Meta is pending. What Florida preserved was its ability to bring another one, subject to the ordinary rules governing jurisdiction, venue, limitations, and transfer. And its attorney general is now publicly saying Florida intends to take Meta to trial.

It is easy to read that as posturing. The arithmetic says otherwise.

Consider New Mexico. It has roughly two point one million residents — about six-tenths of one percent of the country. The settlement allocates money by negotiated schedule, not per capita — and no schedule exists for a state that never joined — but using population as a rough benchmark, New Mexico’s slice of the guaranteed $12.7 billion would have sat somewhere near $80 million. New Mexico instead tried its case, and between the jury’s penalty verdict and the court’s abatement order it stands to recover on the order of $942 million — roughly twelve times that population benchmark. Those judgments are on appeal and the final figure may change. The order of magnitude is the point.

Texas, which was not a plaintiff in the federal case and settled its own claims separately, secured more than a billion dollars for itself while obtaining substantially similar core platform commitments — the two-hour teen limit, school-hours notification blackout, hidden like counts, stricter age assurance.

Florida has more than twenty-three million residents, and the same population benchmark comes to roughly nine hundred million dollars of the guaranteed amount — spread over ten years, with a further slice hostage to whether two competitors decide to cooperate. Set against what a state willing to try its case has actually recovered, walking away is not a gesture. It is a reasoned bet on leverage, and the evidence available today supports it.

There is a second advantage, and it is the one lawyers will notice first.

The consent judgment provides that it does not constitute an admission of liability by Meta, and — in the same paragraph — that nothing in it shall be construed to “apply to, establish a standard of care for, or serve as precedent in any non-participating U.S. state.” A separate provision records the parties’ intent that the judgment not be admissible in other cases against Meta, while expressly preserving Meta’s right to introduce it in disputes over its own insurance coverage or to enforce the releases it bought. Meta agreed to protect children and simultaneously secured the states’ agreement that doing so proves nothing about whether it should have protected them all along.

Florida is a non-participating state. It released none of its claims and accepted none of the settlement’s payment structure. That does not make the settlement itself affirmative evidence in a future Florida case — Rule 408 and the judgment’s own nonprecedent language remain — but Florida preserved its ability to litigate its claims on the underlying evidentiary record. If Florida refiles, it can litigate that case without having exchanged its claims for a payment structure that is thirty percent contingent on the conduct of other companies.

The risk is real and should be named: a promised case still has to be filed, a trial can be lost, and an appeal can take years. But the states that pushed their cases furthest have produced some of the largest individual-state outcomes, and Florida brings eleven times the population of New Mexico to the same fight.

What this settlement does not touch

Here the document turns decisively in favor of families, and it does so in language that leaves nothing to argue about.

The release runs only from the attorneys general, on behalf of their states. It expressly reserves and does not release:

“Any claims of private individuals for any types of monetary or injunctive relief…”

“Any claims, other than the Actions, by plaintiffs, including school district, school, municipality, township, tribal, political subdivision or any other governmental unit plaintiffs, whose lawsuit is pending or subject to transfer to…” the California coordinated proceeding, the federal multidistrict litigation, “or other state and federal courts in the United States.”

Thousands of those claims are live right now — individual young people and their families in the federal multidistrict litigation, and hundreds of school districts. Counsel for those plaintiffs said within hours that this settlement resolves nothing of theirs. The agreement itself says the same.

What the states settled were the states’ claims. A family’s claim was never the states’ to give away, and it was not given away.

Florida’s own line

Section 501.1736 of the Florida Statutes — passed as House Bill 3 — bars social media accounts for children under fourteen and requires a parent’s consent at fourteen and fifteen. A federal judge preliminarily enjoined it on First Amendment grounds in June 2025. In November the Eleventh Circuit stayed that injunction, two to one, concluding the law is likely content-neutral and likely survives intermediate scrutiny — and allowing Florida to enforce it while the appeal proceeds. Argument was heard in March, and no decision has issued as of this writing. The constitutional question is genuinely open and the industry is pressing it hard.

So a Florida parent occupies an unusual position this week: outside a national settlement, inside a state law being enforced but not finally upheld, with an attorney general who turned down the check and promised a trial, and holding a claim no one has settled.

What to take from this

Twenty years of consumer safety law follow one pattern. A product arrives, scales faster than anyone can measure what it does, and is met years later by enforcement negotiated between institutions with every reason to draw the boundaries narrowly. The people actually harmed are rarely in the room where those lines get drawn.

What is unusual here is how much of it is on the record. Meta’s own researchers wrote down what they found. A federal judge has already ruled that Meta had not followed COPPA’s notice and parental-consent requirements — while leaving whether the statute applied to Meta for trial — and said Meta’s denial strained credulity. Two juries have returned verdicts. And the settlement itself, in writing, leaves private claims exactly where it found them.

For a parent who believes a child was harmed, the practical consequence is this: nothing that happened in Oakland on August 26 closed your door. If anything, the record built to get there is now part of the landscape — and the independent auditor the agreement installs must publish an executive summary of its findings each year, even though the full reports stay confidential. Account histories, usage data, school records, and the timeline of when treatment began are what matter, and they are far easier to preserve now than to reconstruct in two years.

The headline said a matter was resolved. The document says something more precise and considerably less final. In this state, the difference between the two is not a technicality. It is the whole of what families still have.

Carter W. Scott is a shareholder of Searcy Denney Scarola Barnhart & Shipley, P.A. He is licensed to practice in Florida and Georgia and has been admitted pro hac vice in courts across the country. From the firm’s Tallahassee office, he tries catastrophic injury and wrongful death cases against product manufacturers, trucking companies, hospitals, and their insurers. This article is general information about a court settlement and pending litigation. It is not legal advice, and it does not create an attorney-client relationship.