Meta has agreed to pay up to US$17 billion over 10 years to settle claims brought by a bipartisan coalition of state attorneys general. The states argued that the company deliberately designed Facebook and Instagram to hook children into using its apps, misled the public about the harm and improperly collected data from children under 13.
The settlement, announced by Meta on Aug. 26, 2026, ended a federal trial that had barely begun in Oakland, California. The potential financial exposure in the case was enormous, and Meta’s stock price took a hit. The states argued that penalties could reach hundreds of billions of dollars.
Against that backdrop, and with a $1.4 trillion question mark hanging over its valuation, Meta settled, although the company continues to deny wrongdoing. The settlement still requires approval from Judge Yvonne Gonzalez Rogers.
As a technology policy and law scholar, I follow with interest the litigation against Meta and other social media companies. The basic contours of this settlement are now public, and I believe they deserve careful attention because of the product design changes it might compel Meta, TikTok and YouTube to make.
The Oakland case consolidated lawsuits filed in 2023 by 29 state attorneys general following a nationwide investigation that began in 2021.
Similar to prior litigation in Los Angeles and New Mexico, the states alleged that Meta engineered features – for example, infinite scroll, autoplay, push notifications, likes and appearance-altering filters – to exploit vulnerabilities associated with adolescent development and to maximize engagement.
Understand how AI is changing society, with our weekly newsletterSubscribe for freeThis site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
States alleged that Meta’s internal research documented links between Instagram use and harms including depression, anxiety and body-image concerns among young people, while the company publicly minimized or concealed those risks.
A third set of claims concerned Meta’s collection of data from children under 13 without parental consent, allegedly in violation of federal children’s privacy law. And the states argued that these practices violated state consumer protection statutes as well.
That consumer protection claim is particularly important. For decades, Section 230 of the Communications Decency Act has protected platforms from liability for content posted by their users. But the states sued Meta instead over the company’s own product design, business practices and alleged misrepresentations, not user-generated content.
In opening arguments, California’s lawyer compressed the theory into four words, noting Meta’s business model was to “hook” users, “hold” them, “harvest” their data and “hide” the harm. And there was already precedent at the state level, in Massachusetts, of courts allowing this kind of claim to proceed.
It is important not to conflate the cases. While the federal case in Oakland was settled as part of the agreement with the 47 states, major cases against Meta remain active.
In Los Angeles, a separate California state-court case resulted in a jury finding Meta and Google liable for negligently designing their products in ways that contributed to a teen girl’s mental health harms, with damages of $4.2 million against Meta and $1.8 million against Google. Those amounts look modest until multiplied across the thousands of similar individual claims now pending. Meta is appealing that decision.
And in New Mexico, the state pursued its own enforcement action, alleging that Meta endangered children and violated state consumer protection law. New Mexico won judgments against Meta totaling more than $900 million, which Meta is also challenging.
Neither of those cases was part of the federal trial, so today’s settlement does not resolve them, nor does it create legal precedent.
The design changes, such as curbing infinite scroll and autoplay, are the substantive core of the agreement. Subject to court approval, teens under 18 on Instagram and Facebook in participating states will get:
a default two-hour daily time limit, cumulative across both apps and across multiple accounts, that only a parent can lift
a block on app access between midnight and 6 a.m. and muted notifications during school hours (8 a.m. to 3 p.m.), excepting direct messages
usage prompts after every 15 minutes of continuous scrolling
the option of a non-algorithmic, nonpersonalized feed
the ability to turn off autoplay and for parents to set the default to “off”
hidden like counts by default
blocks on cosmetic surgery and extreme makeup filters
strengthened age-detection systems for under-13 accounts
Some of these measures are particularly significant because they change the default experience rather than simply adding another setting that users can choose to activate. This lowers the burden on the user. A safety tool that requires a teenager or parent to find, understand and turn on is fundamentally different from a safety constraint built into the product itself.
The agreement addresses the architecture that determines how the product operates. In doing so, it recognizes that Meta shares responsibility for the environment it creates, which is crucial because Meta holds significant power to shape how its products are used.
Even $17 billion, spread across a decade, amounts to only roughly 1% of Meta’s expected revenue over the same period. The company has told investors that the settlement will not change its financial guidance beyond a single quarterly expense.
More consequential, for Meta and the public, is the requirement to redesign Facebook and Instagram.
However, redesign without serious transparency and accountability can quickly become another form of marketing. Meta has promised safety changes before, including teen accounts and expanded parental controls. Verification of the redesign and its real-world impact over time are key.
That is where the settlement is both promising and incomplete. An independent auditor will review Meta’s compliance annually, but only for five years, against a 10-year agreement. And Meta has committed to the strongest behavioral terms, the daily time limit and the overnight block, for just five years. It will extend to 10 with stricter defaults only if YouTube and TikTok sign on.
Meta has also committed to establish an independent research foundation. And this might address one of the deepest problems in technology governance: Platforms hold the evidence of their own effects, while independent researchers have been locked out.
But the strength of these mechanisms will depend on details not yet public. Who selects the auditor? What information can the auditor access? Can researchers independently reproduce findings? What constitutes noncompliance, and what penalties follow if Meta complies with the letter of the agreement while redesigning the product around its edges? Finally, how does the research foundation escape capture by Meta?
Technology governance increasingly fails not at the level of rules but at the level of enforcement.
There is also something the settlement forecloses. Trials produce public records, but settlements end the process of producing a record of the evidence. The internal documents and testimony surfacing in Oakland, including evidence about Meta’s own research on young users and its lawyers’ handling of those findings, will now remain only partially visible. The states secured a significant legal resolution, but the public lost part of the record of how we got here.
Three things deserve close attention:
First, the implementation details matter.
The settlement’s significance depends on what happens after the press releases disappear. Do the new defaults actually constrain use? Are they hard to circumvent across accounts and devices? It is very hard to do age control in technology. And will teenagers simply migrate to other platforms outside the settlement’s reach, or into the direct messages the agreement itself exempts?
Second, there is the question of Meta’s competitors.
Participating states are guaranteed roughly 70% of the $17 billion, or about $12.7 billion over the decade, while the remaining 30%, about $5.3 billion, is released only if YouTube and TikTok adopt comparable teen protections and make matching payments. Meta essentially structured part of its own penalty as a recruitment device, effectively telling TikTok and YouTube that these rules are coming, and it is better to adopt them together than face the next litigation cycle alone.
There is also a clear economic incentive for Meta to try to level the playing field. Because if these measures make Meta’s products less profitable, imposing similar requirements on competitors could prevent Meta from bearing those costs alone. Whether that strategy works could matter as much as the settlement itself.
Third, federal social media policy in the United States is gridlocked. It took state attorneys general, applying consumer protection law, to produce rules about time limits, product defaults, notifications and design practices that Congress could have debated and enacted years ago.
The deeper lesson of today’s settlement is that product design, as a site of legal accountability, is a workable legal strategy.
Accountability starts when a company agrees to redesign its product. The settlement gives the states – and the rest of us – a decade to find out whether these changes are real, and if so, if they work for the desired policy outcome.
The critical question is whether, throughout that decade, someone independent will have the authority, access and incentives to decide what “work” really means.