How Meta Learned Not to Know: Meta (Facebook) $1.4 Trillion Lawsuit:
Some numbers stop being numbers and become arguments. One point four trillion dollars: that, by Meta’s own calculation in a court filing dated July 6, is what the penalties would come to if California, Colorado, Kentucky, and New Jersey win at trial this August and their method of counting violations carries the day. The figure sits uncomfortably close to the company’s entire market value. Meta’s lawyers called the proposals “outlandish” and observed that a sanction of that size has never been seen in the history of consumer protection enforcement. On this narrow point they are correct. But then nothing about this case has been seen before, and to understand where a sum like that comes from, you have to go back to a meeting that never took place.
It was set for April 2, 2020, and it was supposed to settle the fate of Instagram’s camera filters that simulated plastic surgery. The briefing memo summarized consultations with twenty-one independent experts: effects of this kind can do serious damage to mental health, children are especially vulnerable, and the long-term studies would not exist until well after the harm had materialized. The same memo, with a bookkeeper’s candor, noted that keeping the filters banned might have a “negative growth impact.” The meeting was cancelled on April 1, the day before it was to be held; the record does not indicate that anyone appreciated the date. No new meeting was scheduled. Mark Zuckerberg vetoed the ban by correspondence. He dismissed the objections of experts, the public, and his own employees as “paternalistic,” and said he had seen no data showing the filters were harmful. The data sat in the memo he had been given.
Margaret Gould Stewart, then Meta’s Vice-President of Product Design and Responsible Innovation, wrote back to her chief executive with a sentence that the state attorneys general would later make one of the pillars of their complaint. She would support the decision, she said, but wanted it on the record that “I don’t think it’s the right call given the risks.” She announced her departure in October of 2022 and left the following month. In October of 2023, as the complaint was being filed, the filters were still live.
This is the third installment in a series on the Meta litigation. The undercover operation in Santa Fe was the subject of a February piece, New Mexico v. Meta & Facebook; the three-front legal war and the fate of Section 230 were covered in March, in Facebook on Trial. Today’s subject is a pair of documents best read together: the less-redacted version of the multistate complaint, unsealed on November 22, 2023, in the Northern District of California (233 pages, Case No. 4:23-cv-05448-YGR), and the transcript of the whistleblower hearing held before a Senate Judiciary subcommittee on September 9, 2025 (S. Hrg. 119-255, “Hidden Harms”). The first document describes how a corporation hid what it knew. The second describes the next stage: how it learned not to know. The law has an old name for that condition, and an expanding set of tools.
The Company of One Veto
What makes the filter episode instructive is not its ending but the decision architecture it exposes. Every layer of the organization did its job. Researchers gathered the data; the experts were invited and heard; a vice-president made her recommendation; the heads of Instagram and Facebook sat on the email thread. The system stopped at a single person, who needed only to cancel a meeting and write that he had seen no data. The complaint documents the same choreography around other decisions, and the filter story eventually resurfaced in a Los Angeles courtroom, where Zuckerberg confirmed under oath both the experts’ recommendation and his refusal.
The 2025 testimony adds an epilogue to that architecture. After Frances Haugen’s disclosures, the so-called Zuck reviews, formal project readouts with the chief executive in the room, quietly vanished from the calendar. The whistleblowers were blunt about the purpose: no written trail of what he knew. Organizational psychology has a name for the mechanism, strategic ignorance: inconvenient information is not fought but routed around, so that it never formally reaches the person in charge. Corporate language calls it plausible deniability. Evidentiary language calls it the plaintiff’s burden of proving actual knowledge. For a while, anyway; more on that below.
The Budget as Testimony
Corporate declarations are cheap. Budgets testify under oath. In August of 2021, product teams wrote to Chris Cox and Nick Clegg: after reviewing more than a thousand well-being topics, they had settled on four priorities (problematic use, bullying and harassment, social connection, and content about suicide and self-injury), the company was on track to succeed at none of them, and there was no central ownership and almost no staff. Clegg forwarded the request to Zuckerberg with a recommendation to invest, citing pressure from politicians and a survey of American policy elites. Cox, the chief product officer, sized up the prospects in an internal message: “very low-likelihood that Mark chooses to fund more here.” Months of silence followed. In November of 2021, Clegg renewed the request, already scaled back, arguing that the investment was needed so the company could “stand behind our external narrative of well-being” in its apps and, soon, in the metaverse. Naomi Gleit endorsed it as her top project below the line. Susan Li deferred to Mark. The fiscal outcome, recorded by Vice-President Emily Dalton Smith, was zero new well-being funding for 2022.
To an analyst, this is the most valuable stretch of the unsealed material, not because it is dramatic but because it is accounting. An economist would call it revealed preference: a firm that publicly declares child safety its priority and internally funds it at zero has made a statement about its hierarchy of goals more precise than any press release. Prosecutors call the same thing evidence of intent. And Clegg’s phrase about the external narrative performs work its author surely never intended: it confirms, in writing, that in this organization the narrative and the product were two separate line items, each requiring its own budget.
A North Star Not for Showing
An internal research memo describes the TRIPS survey, the Tracking Reach of Integrity Problems Survey, as “our north star, ground-truth measurement.” The company navigated by it carefully. A TRIPS report dated May 11, 2020, showed that 19.3 percent of Instagram users had encountered hate speech in the survey window, 12.2 percent had seen graphic violence, and more than 20 percent had witnessed bullying. A separate 2018 study estimated that 5.1 million Instagram users encounter suicide and self-injury content every day. In public, quarter after quarter, the Community Standards Enforcement Reports cited prevalence figures measured in hundredths of one percent of views (0.05 percent, for some categories of harm). The mechanics of that gap, the difference between measuring people exposed and measuring the share of views, were dissected in the March article; here the complaint’s conclusion will do. One metric was treated as the truth about the product. The other was the truth for the public.
Four Million Children and the Word “Penetration”
Count One of the complaint, joined by every filing state, arises under COPPA, the Children’s Online Privacy Protection Act (15 U.S.C. § 6501 et seq. and 16 C.F.R. Part 312): a federal prohibition on collecting personal data from children under thirteen without verifiable parental consent. The platforms’ defense has been the same for years. The terms of service forbid registration under thirteen; therefore the company has no actual knowledge of children among its users. The unsealed passages dismantle that construction in bursts. In January of 2018, Zuckerberg received a report estimating that in 2015 Instagram had roughly four million users under the age of thirteen, about 30 percent of all American children between ten and twelve. Internal charts tracked the activity of the eleven-to-thirteen cohorts, reaching from 20 to 60 percent of a birth-year cohort in monthly terms, and described it with the word “penetration,” a term from the vocabulary of market conquest, not violation removal. In July of 2021, a “youth dashboard” circulated with engagement broken out into U13, thirteen to seventeen, and eighteen plus.
The legal punch line has a name and a citation. The Supreme Court set out the willful blindness doctrine in Global-Tech Appliances, Inc. v. SEB S.A., 563 U.S. 754 (2011), in two elements: a subjective belief that a fact is highly probable, and deliberate actions taken to avoid confirming it. The doctrine was minted in patent law, and its transfer to the consumer protection context is the states’ argument rather than settled appellate authority; even so, it is hard to imagine a cleaner textbook illustration than an organization that measures the “penetration” of age cohorts it officially does not have, while designing its sign-up flow so as not to ask about age too effectively. Two procedural morsels for connoisseurs: Maine, alone among the states, declined to join the monetary demands under COPPA, and the state statutes count penalties per violation, up to ten thousand dollars in Arizona. At this scale of users, the arithmetic explains the legal department’s nerves better than any theory. It also explains how, years later, four states could arrive at a demand with twelve zeros in it.
Reality Labs: So the Data Never Exists
The complaint describes the concealment of knowledge that existed. The testimony given on September 9, 2025, by Dr. Jason Sattizahn and Cayce Savage, two of six researchers covered by a disclosure coordinated through Whistleblower Aid, describes the next stage, rolled out after 2021 in the company’s virtual-reality division: preventing the inconvenient knowledge from ever coming into being. The witnesses called it a funnel of manipulation. Lawyers supervise the design of studies, restrict what data may be collected, edit the reports, and silo the results so that no one sees the whole.
The most vivid episode comes from Germany. During a research interview, a teen-ager related that his younger brother had been sexually propositioned in VR. The mother was present and had given her express consent to the children’s participation. The legal department ordered the recording and all notes deleted. Kristin Zobel, the lawyer relaying the instruction, reportedly added, “I know, this is ridiculous, but we have to do it.” Another researcher was given to understand that he would presumably not want to testify in public if the findings ever got out. Outside vendors, the witnesses said, were engaged partly so that data could be erased and the company could maintain that it held none. Project Horton, an age-verification program worth more than a million dollars and approved at the C.T.O. level, was cancelled without explanation; only one person in the organization outranks the C.T.O. Project Salsa, meanwhile, lowered the minimum age for the headsets from thirteen to ten, in the form of parent-managed accounts; in documents cited at the hearing it figured as an “alternative” response to F.T.C. pressure, and the logic is disarmingly simple. A legalized ten-year-old is a ten-year-old you no longer have to detect and remove. Estimates quoted at the hearing put the share of under-age users in some VR spaces at 80 to 90 percent, and in research Sattizahn cited, 23 percent of women reported verbal sexual advances, against 9 percent of men. Meta rejects the allegations; the company has dismissed the claims at the heart of the hearing as nonsense.
Lawyers in the Laboratory
The sharpest legal thread of the hearing concerns the role of counsel. The witnesses described instructions to avoid the words “illegal” and “noncompliant” in research documents, and the routine copying of lawyers on correspondence so that privilege would attach. Senator Ashley Moody, a former attorney general of Florida, put the matter plainly: privilege protects legal advice, not participation in perpetuating a wrong. American doctrine has long known the crime-fraud exception, which strips protection from communications made in furtherance of ongoing wrongdoing; the Supreme Court shaped the in camera procedure for testing it in United States v. Zolin, 491 U.S. 554 (1989). The second blade is spoliation. Since the 2015 amendments, Rule 37(e) of the Federal Rules of Civil Procedure allows a court, where a party intentionally deprives an opponent of electronic evidence, to go as far as an adverse-inference instruction to the jury. An order to delete recordings of research whose subject is harm to a child, issued on the eve of entirely foreseeable litigation, is the scenario the rule was written for. A marginal note for the Polish reader: here, too, professional privilege protects legal assistance, not a lawyer’s participation in wrongdoing. The line runs in the same place; it is simply tested less often under this kind of pressure.
Europe Outlaws Ignorance: the DSA and the GDPR
The American dispute is about what the company knew. The European legislator inverted the question and made ignorance itself the violation. The Digital Services Act (Regulation 2022/2065) requires, in Article 34, that very large platforms conduct an annual assessment of systemic risks, expressly naming the protection of minors and mental health, and, in Article 35, that they adopt mitigation measures. A firm that does not know what its product does to children has no defense in this regime, because the duty to know comes first. Brussels also has a provision for deleted research. Article 40 of the DSA grants vetted researchers access to platform data, which means knowledge about the product can no longer be kept exclusively in-house. Article 28 adds a ban on profiled advertising to minors. One caveat for the Polish reader: the regulation binds directly, yet Poland has still not enacted the statute standing up its national enforcement apparatus. The Sejm passed an implementing law in December of 2025, the President vetoed it, and Poland remains the only member state without a designated Digital Services Coordinator; the duty to know is already in force, the domestic machinery to police it is not. In the background sits the GDPR, whose Article 8 lets member states set the age of a child’s independent consent between thirteen and sixteen; Poland declined the option to lower it, so the threshold there is sixteen.
State of Play: July 2026
Since March, the landscape has changed fundamentally. On March 24, a jury in Santa Fe found Meta liable on both counts under New Mexico’s Unfair Practices Act, found the conduct willful, and imposed the statutory maximum of five thousand dollars per violation, a total of three hundred and seventy-five million dollars. Midway through the jury trial, Meta announced it would withdraw end-to-end encryption from Instagram messages; the change took effect on May 8. A second, bench-trial phase on public nuisance opened before Judge Biedscheid on May 4, with the state seeking sums approaching a billion and product changes led by age verification. A day later, on March 25, a Los Angeles jury found Meta and YouTube liable to the twenty-year-old plaintiff known as Kaley: three million dollars in compensatory damages, split 70 to 30, plus a further three million in punitive damages ($2.1 million from Meta, $900,000 from YouTube) after a finding of malice, oppression, or fraud. Both companies are appealing; a second bellwether and the first federal trial of the school districts’ claims are set for this summer. Meta’s appeal in the Ninth Circuit (No. 24-7032) is heading, after the January 6 argument, toward a purely jurisdictional resolution: the panel signaled that Section 230 confers immunity from liability, not from suit, which would send the states’ case to the merits before Judge Gonzalez Rogers, with the Section 230 question deferred to an appeal from final judgment.
The federal front accelerated at the turn of June and July. On June 30, Judge Gonzalez Rogers denied Meta’s bid to resolve the case without trial, holding that whether the platforms were designed to be addictive, whether the company falsely denied it, and whether they were even partly directed to children remain disputes of fact; if the plaintiffs’ evidence bears that out, she wrote, a jury could reasonably find the company’s public assurances untrue. Trial is set for August 18 in Oakland. It will take up the COPPA claims now pressed by twenty-nine states, together with the consumer claims of four of them, California, Colorado, Kentucky, and New Jersey; a further fourteen states await a separate trial in February. The stakes were defined, paradoxically, by the defendant itself: in the July 6 filing, Meta priced the four states’ penalty methodology, statutory per-violation fines multiplied by the estimated number of affected minors, at $1.4 trillion, accusing the plaintiffs of counting the same teen-agers two and three times over and of exceeding the limits of due process. A sanction of that size, the defense added, “has no analog in the history of consumer protection enforcement.” True. The trouble is that neither does the evidentiary record described above.
Return, finally, to the April, 2020, e-mail. Margaret Gould Stewart, acquiescing in the veto, expressed the hope that in the years to come the company would look back and feel good about the decision. Six years have passed. Three bodies have so far answered her question: a jury in Santa Fe, a jury in Los Angeles, and a committee of the United States Senate. Each in its own procedure. None in the affirmative. The fourth convenes on August 18, in a courtroom in Oakland.
Robert Nogacki Attorney at Law (radca prawny, WA-9026), Managing Partner, Kancelaria Prawna Skarbiec, Warsaw This article reflects the author’s personal views and does not constitute legal advice.
Sources:
Complaint for Injunctive and Other Relief (less-redacted version, Dkt. 73-2), Case 4:23-cv-05448-YGR (N.D. Cal., Nov. 22, 2023) | Hidden Harms: Examining Whistleblower Allegations that Meta Buried Child Safety Research, S. Hrg. 119-255 (Senate Judiciary Subcommittee, Sept. 9, 2025) | Global-Tech Appliances, Inc. v. SEB S.A., 563 U.S. 754 (2011) | United States v. Zolin, 491 U.S. 554 (1989) | Fed. R. Civ. P. 37(e) | Regulation (EU) 2022/2065 (DSA), Arts. 28, 34, 35, 40 | GDPR, Art. 8 | NMDOJ press release on the verdict (Mar. 24, 2026) | Source New Mexico on phase two | NPR on the Los Angeles verdict (Mar. 25, 2026) | CNN on the verdict and trial evidence | Reuters (via Claims Journal) on Meta’s July 6 filing and the $1.4 trillion figure (July 7, 2026) | The Epoch Times on the June 30 denial of summary judgment | EPIC on the Ninth Circuit argument (Jan. 7, 2026) | Reuters on the $1.4 trillion filing (July 7, 2026) | Washington Post investigation on suppressed VR research (Sept. 8, 2025) | Business Insider on the unsealed TRIPS figures (Nov. 2023) | NPR on Zuckerberg’s February testimony (Feb. 18, 2026) | Reuters on Gould Stewart’s departure (Oct. 14, 2022) | BBC on Meta’s response to the hearing | TechCrunch on the Quest age change (June 2023) | Security Affairs on the Instagram encryption removal (May 2026) | Łakomiec & Grochowski, I·CONnect, on Poland’s stalled DSA implementation (May 7, 2026)
Robert Nogacki – licensed legal counsel (radca prawny, WA-9026), Founder of Kancelaria Prawna Skarbiec.
There are lawyers who practice law. And there are those who deal with problems for which the law has no ready answer. For over twenty years, Kancelaria Skarbiec has worked at the intersection of tax law, corporate structures, and the deeply human reluctance to give the state more than the state is owed. We advise entrepreneurs from over a dozen countries – from those on the Forbes list to those whose bank account was just seized by the tax authority and who do not know what to do tomorrow morning.
One of the most frequently cited experts on tax law in Polish media – he writes for Rzeczpospolita, Dziennik Gazeta Prawna, and Parkiet not because it looks good on a résumé, but because certain things cannot be explained in a court filing and someone needs to say them out loud. Author of AI Decoding Satoshi Nakamoto: Artificial Intelligence on the Trail of Bitcoin’s Creator. Co-author of the award-winning book Bezpieczeństwo współczesnej firmy (Security of a Modern Company).
Kancelaria Skarbiec holds top positions in the tax law firm rankings of Dziennik Gazeta Prawna. Four-time winner of the European Medal, recipient of the title International Tax Planning Law Firm of the Year in Poland.
He specializes in tax disputes with fiscal authorities, international tax planning, crypto-asset regulation, and asset protection. Since 2006, he has led the WGI case – one of the longest-running criminal proceedings in the history of the Polish financial market – because there are things you do not leave half-done, even if they take two decades. He believes the law is too serious to be treated only seriously – and that the best legal advice is the kind that ensures the client never has to stand before a court.
