Settlement Without Admission: Meta's Legal Reckoning Over Children, Algorithms and Platform Accountability

The settlement that avoided a verdict

In August 2026, Meta Platforms reached a reported multibillion-dollar settlement with 29 American states after roughly a week of trial.¹ The company admitted no liability. The central allegations in the multistate action, that Meta knowingly designed Facebook and Instagram in ways that harmed children and concealed what it knew, were therefore not tested to a final jury verdict in that proceeding.¹

What the settlement bought was not vindication. It bought finality, speed and containment. It ended a public trial that threatened to expose internal decision-making about children, engagement, product design, safety investment and executive accountability.

A record already forming

The settlement did not arrive in isolation. Before it, juries and courts had already begun to test parts of the broader case against Meta. In New Mexico, a jury found that the company had violated state consumer protection law in connection with youth safety claims.² In California, a Los Angeles jury in a bellwether social media case found Meta and YouTube negligent and liable for failure to warn, while also finding malice, oppression or fraud.⁴ Meta has disputed adverse findings and indicated it will appeal.

Meta admitted no liability in the August 2026 settlement and says it will appeal both jury verdicts. The states' central allegation has never been put to a jury in the multistate case.

Source: Wall Street Journal; US Senate Judiciary Subcommittee on Privacy, Technology and the Law; State of New Mexico v Meta Platforms; JCCP 5255, Superior Court of California; FTC v Meta Platforms; Northern District of California.

NOFUF

The company has also had significant legal wins. In 2025, a federal judge rejected the Federal Trade Commission's monopolisation case, finding that the regulator had not proved Meta held monopoly power once competitors such as TikTok and YouTube were considered.⁶ That ruling left Instagram and WhatsApp inside Meta's corporate structure.

Warnings from inside

The public record includes years of internal warnings and former employee testimony. In 2021, former Facebook product manager Frances Haugen provided internal documents to the Wall Street Journal and filed disclosures with the US Securities and Exchange Commission.⁷ Among the materials reported publicly were internal findings about teenage users, body image and Instagram's effect on some young people.⁸

That same year, former Facebook engineering director Arturo Bejar, who had returned as a consultant, emailed senior executives including Mark Zuckerberg and Adam Mosseri with survey data about teenage users' experiences. Later, testifying before the US Senate, Bejar said his recommendations were not adopted.⁹ His evidence placed the issue in stark terms: the company had been told, in detail, that young users were encountering unwanted sexual advances, bullying and negative social comparison on its platforms.

In 2025, former Meta researchers made further allegations to a Senate subcommittee about child safety research in virtual reality products, including claims that findings were suppressed or altered.¹⁰ Meta rejected the characterisation, saying the allegations reflected a selective and false narrative and that data removals complied with privacy obligations relating to under-13 users.

The algorithm question

The allegation advanced by whistleblowers, regulators and critics is not simply that harmful content appears on Meta's platforms. It is that engagement-based ranking systems can elevate the material most likely to provoke reaction, and that outrage is among the most reliable forms of engagement.

Internal documents reported in 2021 indicated that Facebook's 2018 shift toward so-called meaningful social interactions, including heavier weighting of reactions, was found internally to amplify divisive material.⁷ Meta has disputed the broader interpretation of those documents and has pointed to later ranking and safety changes.

The research record is contested. Studies conducted with Meta's cooperation around the 2020 US election found that chronological feeds did not significantly change political polarisation over the study period.¹⁵ But later editorial concern noted that the findings had limits, including the fact that Meta had already introduced emergency measures reducing some political content during the period studied.¹⁶

When automation gets it wrong

Beyond the courtroom lies a quieter but more ordinary problem: account enforcement. In 2025, users around the world reported wrongful suspensions across Facebook and Instagram, including bans under severe policy categories.¹⁸ Meta acknowledged some technical errors, including issues affecting Facebook Groups. For users whose accounts disappeared, the question was not only whether automated systems can make mistakes. It was whether the company provided a meaningful path to correct them.

Meta's own Oversight Board has criticised aspects of the company's approach to permanent account bans, including concerns about transparency, access to appeal and meaningful human review.¹⁷ Those findings go to the heart of the platform's power: when a user loses an account, they may lose a business, an audience, years of personal records and access to communities, often without a clear explanation from a person empowered to fix the decision.

Support as a paid product

Meta Verified launched in 2023, with Australia and New Zealand among the first markets. One of the subscription's practical attractions is access to enhanced support. For individuals and businesses, that raises a basic fairness issue: if automated systems can remove an account that functions as a livelihood, should access to meaningful human help depend on whether the user pays for a premium product?

The scale argument is familiar. Meta serves billions of people; no system of human support could examine every dispute in detail. But scale is not an answer to every design choice. Other large technology companies operate at enormous user or device scale while maintaining free human support channels for at least some categories of customer problem. The issue is not whether Meta can eliminate error. It is whether the company has invested enough in remedy when its own systems cause harm.

Meta's most-quoted figure is more than US$5 billion a year, a number it first used for 2021. Set against 2025 revenue it is about 2.5 per cent. Meta has not published an updated annual figure.

Source: Meta Platforms Form 10-K, financial year 2025; Meta public statements on safety and security spending.

NOFUF

Power without ordinary shareholder control

Meta's governance structure concentrates voting power in Mark Zuckerberg's hands through its dual-class share system. Class B shares carry ten votes each, while Class A shares carry one.²¹ As long as that structure remains in place, outside shareholders cannot force fundamental change without Zuckerberg's support.

Class B shares carry ten votes each and Class A one. He holds about 99.7 per cent of the Class B stock. No combination of outside shareholders can outvote him.

Source: Meta Platforms proxy statement and Form 10-K, financial year 2025.

NOFUF

That matters because the questions raised by the litigation, whistleblower disclosures and enforcement failures are not merely operational. They are questions of corporate priority: how much friction to place in front of growth, how much money to spend on safety, how much access ordinary users should have to appeal, and how much accountability should sit with the person who controls the company.

The Australian test

Australia's minimum-age regime has put those questions into a sharper local frame. The law requires age-restricted platforms to take reasonable steps to prevent under-16s from holding accounts, with substantial penalties for non-compliance.²⁸ Meta criticised the legislation as rushed, yet later reported the removal of hundreds of thousands of Facebook and Instagram accounts in Australia assessed as belonging to under-16s.²⁷

Those accounts existed on Meta's platforms despite years of assurances that underage users were not permitted. That fact does not resolve the policy debate about age restrictions. It does, however, expose the gap between platform rules, platform enforcement and platform reality.

What remains unanswered

The unresolved issue is not whether Meta has built useful products. Billions of people use them because they are useful. The issue is whether a company that mediates communication, commerce, identity, childhood and political information at global scale can continue to treat meaningful accountability as optional, litigated only after damage is done and sold as a premium service when ordinary users need help.

Meta paid to end one major case without admitting wrongdoing. It did not answer the larger question. If the company's systems can shape what children see, what adults believe and whether users keep access to their own accounts, then the public interest lies not only in what Meta says about safety, but in what it can be made to prove.