New Jersey is assured about $545.9 million across two settlements and could receive up to $772.5 million, but the agreement still requires a federal judge’s approval.
MORRISTOWN, NJ – Meta has agreed to pay at least $12.1 billion and potentially $17.1 billion to states and U.S. territories while imposing new restrictions on young people’s use of Facebook and Instagram, ending a federal trial co-led by New Jersey before a verdict was reached.
New Jersey is assured approximately $545.9 million if the proposed judgment takes effect. That consists of about $525.6 million from the youth social media case and $20.3 million from a separate Cambridge Analytica privacy settlement. Another $226.6 million is conditional, bringing the state’s possible total to approximately $772.5 million.
The settlement is not yet final. The states and Meta filed a joint motion asking U.S. District Judge Yvonne Gonzalez Rogers to enter the agreement. Until that happens, Meta’s payments and compliance deadlines do not begin.
The New Jersey Attorney General’s Office described the package as the largest multistate attorney general settlement in U.S. history.
Where the $17.1 billion figure comes from
The 130-page proposed consent judgment separates the money into several parts.
Its payment schedule lists approximately $11.66 billion in guaranteed youth-case payments over 10 years, plus about $5.02 billion in conditional payments. That produces a maximum youth-case figure of approximately $16.68 billion.
A separate $459.3 million agreement resolves state claims connected to the Cambridge Analytica privacy scandal. Added to the youth-case payments, those figures round to a minimum of $12.1 billion and a maximum of $17.1 billion. Meta would also provide $75 million to reimburse participating states for investigation and litigation costs.
Meta has described the broader agreement as approximately $18 billion following Texas’ participation. The filed New Jersey judgment and the state’s announcement use the $17.1 billion figure. New Jersey’s listed allocation is not affected by that difference.
What New Jersey would receive
New Jersey’s youth-case payment is scheduled as 10 installments of approximately $52.56 million, for a guaranteed total of $525,645,673. The first payment would be due within 30 days after the judgment takes effect, with the remaining installments due each Jan. 15.
The state would also receive $20,260,286.38 from the Cambridge Analytica agreement within 30 days. The New Jersey-specific section of the judgment rounds the combined assured payment to $545,905,959.
New Jersey could receive another $226,554,278.70, but that money is not guaranteed.
To unlock it, Snap, TikTok and YouTube must all become subject to comparable nighttime restrictions, daily limits and age-assurance requirements. Companies in that group with annual profits above $10 billion must also face monetary obligations at least comparable to Meta’s conditional payment.
If those conditions are never met during the agreement’s 10-year term, Meta keeps the conditional money.
The settlement does not create a claims process or provide checks directly to parents, teens or other Facebook and Instagram users. The money is paid to the New Jersey Attorney General.
The attorney general would have broad discretion to use it for consumer protection or privacy enforcement, consumer education, litigation, local consumer assistance, investigation costs, attorneys’ fees or other purposes allowed by state law. The judgment does not earmark a particular amount for youth mental health programs.
“Today’s settlement is a major victory for New Jersey families,” Gov. Mikie Sherrill said. “Big Tech is finally being held accountable for the harm it is causing, especially to our children’s mental health and safety.”
Sherrill said the agreement “will bring more than $500 million to New Jersey to continue the fight.”
A two-hour limit, with important exceptions
Under the agreement, a teen is a user Meta identifies or predicts to be between 13 and 17.
Facebook and Instagram would place those users under a default combined limit of two hours per day. Time would be combined across linked accounts and across both platforms.
A teenager could choose a stricter limit but could not loosen the two-hour setting without approval from a parent linked through Meta’s supervision system. A parent could loosen or remove the limit for one day, one week, 30 days or until the parent changes it back.
The limit would not count time spent in direct messages, account settings or watching qualifying long-form content of at least 22 minutes. Those services would remain available after a teen reaches the daily limit.
The initial two-hour limit would last for five years. If the industry-wide conditions involving Snap, TikTok and YouTube are met, the limit would become 60 minutes on each Meta platform, with a maximum of 120 minutes combined, for the remainder of the 10-year term.
Nighttime and school-hour restrictions
Teen users would be blocked from most Facebook and Instagram features between midnight and 6 a.m. Messages and account settings would remain accessible.
Push notifications would be disabled from 10 p.m. to 7 a.m., except for urgent account-security or platform-integrity notices. A supervising parent could change the default.
If the industry-wide conditions are met, the overnight block would expand to 10 p.m. through 7 a.m.
During school hours, defined as 8 a.m. to 3 p.m. on weekdays between Aug. 15 and June 15, push notifications would be off by default. Messages and security notices are exempt. Parents could adjust those hours or block nearly all Facebook and Instagram functions during the school day.
Meta would also show teens notices after 15 minutes of continuous use and increasingly prominent pauses after 60 and 90 minutes of cumulative use. Those notices encourage a break but do not necessarily force the teen to log out.
Changes to feeds, likes, filters and parental controls
Within several months of the judgment taking effect, Meta would have to make a chronological feed composed of followed or friended accounts reasonably accessible to teens. Meta would periodically remind them about the option.
That chronological feed would not become the automatic default for every teen. Teens could still use Meta’s personalized recommendation systems unless they or a parent choose otherwise.
Teens would also be able to turn off autoplay, but autoplay would not be disabled automatically in every account.
Other requirements include:
- Hiding like and reaction counts from teens by default.
- Disabling filters that imitate cosmetic surgery or extreme makeup.
- Keeping Instagram teen accounts private by default and applying appropriate privacy settings on Facebook.
- Limiting teens’ ability to follow or interact with accounts Meta identifies as age-inappropriate.
- Restricting potentially suspicious adults from finding or contacting teens.
- Giving parents more information about linked and secondary accounts, time spent, contacts and reported accounts.
- Maintaining content policies addressing bullying, sexual content, child exploitation, self-harm, eating disorders, violence, gambling and restricted goods.
For potentially harmful content reported in English or Spanish, Meta must maintain a system designed to issue a decision within six hours in at least 90% of cases. Teens would be able to appeal when Meta finds no violation.
Some of these provisions require Meta to maintain and improve its processes. They do not guarantee that a teen will never encounter harmful content.
Age assurance, not a universal ID requirement
Within one year, Meta would have to apply an age-assurance method to every Facebook and Instagram user in participating jurisdictions.
The methods may include technology licensed from outside companies, ID verification, facial age estimation, Meta’s own prediction systems and reliable age signals supplied by Apple or Google. The judgment does not require every user to upload government identification.
Commercial age-assurance systems must limit the rate at which actual teens are mistakenly classified as adults. The maximum error rate is 10% for users ages 16 and 17 and 3% for users ages 13 through 15.
Meta’s internally developed systems would initially be allowed error rates of 14% and 7%, respectively. By the second year, those rates must fall to 10% and 5%.
The methods must undergo annual accredited testing under real-world conditions, including testing across different demographic groups and safeguards against spoofing. Users must be allowed to appeal an incorrect age classification.
For children under 13, Meta would have six months to strengthen its reporting, account-matching and review procedures. When an account is removed as underage, Meta would review its network to identify other possible under-13 users.
Meta would also develop an under-13 age-prediction model within one year and use it to evaluate all accounts during the second year. The agreement allows Meta to set “reasonably achievable” annual removal targets subject to auditor review. It does not impose a fixed minimum number of underage accounts that must be removed.
Restrictions on children’s data
Data collected specifically to determine a user’s age, along with data from known users under 13, generally must be held only as long as needed for that determination and then placed in a deletion queue.
Meta may retain under-13 data when necessary to develop, train, test or measure the model used to detect underage accounts. That information must be protected using Meta’s highest privacy and security standards.
It cannot be used for advertising, marketing, ad delivery, algorithmic optimization or another unrelated purpose. The data must be stored at the least detailed level that remains useful and encrypted in transit and at rest.
Meta may retain limited metadata needed to prevent people from bypassing the age system. That metadata must be deleted within 90 days after it is no longer needed.
The restrictions do not apply in the same way to a user’s stated birth date, stated age or the final age-classification result.
No direct AI or chatbot restrictions
The filed judgment does not impose content rules, time limits or safety requirements specifically on Meta AI or other chatbots.
In fact, the agreement expressly excludes products or features whose primary function is artificial intelligence, chatbots or interactions with AI from the definition of a new social media platform covered by its expansion provisions.
Direct messaging is also excluded from the daily and nighttime restrictions. That matters because some AI interactions may occur through messaging features.
The data-use provision still prevents Meta from taking under-13 information retained for age detection and repurposing it for unrelated AI training or algorithmic optimization.
The trial had included allegations that Meta collected data from children under 13 without parental consent and used it to train machine-learning and generative-AI systems. The settlement resolves the states’ claims without a ruling on whether that occurred.
Independent auditing and New Jersey enforcement
A bipartisan committee of no more than six attorney general offices would work with Meta to select an independent auditor within 60 days.
Meta would pay the auditor, who could examine relevant nonprivileged internal records, raw and aggregate data, systems and personnel. The auditor would issue five annual reports.
A public executive summary would accompany each final report. Most underlying audit records and company information would remain confidential.
If the auditor identifies a material weakness, Meta would have to prepare a corrective-action plan within 30 days. The auditor could require revisions and monitor implementation.
New Jersey would have legal authority to enforce the judgment. Ordinarily, the state must notify Meta of an alleged violation, allow 30 days for a response and participate in 15 business days of good-faith discussions before seeking court action.
New Jersey could act immediately if it believes a violation presents an urgent threat to public health or safety. The federal court would retain jurisdiction over the judgment.
One provision is more restrictive: enforcement of certain allegedly false or misleading statements about teen safety requires approval from a majority of the multistate committee.
Parents and teens cannot enforce the settlement themselves. The agreement expressly creates no private right of action, leaving enforcement to the participating governments and the court.
What came out during the trial
The federal trial began Aug. 18 in Oakland, California. California, Colorado, Kentucky and New Jersey presented state consumer-protection claims, while 29 states pursued allegations under the federal Children’s Online Privacy Protection Act.
An advisory jury heard the evidence, but Judge Rogers was responsible for deciding liability and any relief.
During opening statements, the states cited internal communications describing “teen time spent” as a company goal and employees comparing Instagram to a drug and themselves to pushers. Meta disputed the states’ interpretation and argued that scientific evidence did not establish a clear causal connection between social media use and poor mental health.
Former Meta safety engineer Arturo Béjar testified that Instagram had followed a “don’t ask, don’t tell” approach to children under 13 and that safety was not adequately considered before Reels launched. Meta challenged his account.
Former Meta data scientist George Volichenko testified that adoption of the company’s optional “Take a Break” and “Quiet Mode” tools was very low and that the well-being team’s purpose included protecting Meta from future lawsuits.
Instagram head Adam Mosseri testified that there was “no silver bullet” for teen safety and defended the company’s work. Evidence presented at trial showed that 1.8% of teens adopted the optional Take a Break feature before Meta later made it a default setting.
The settlement was announced after four days of testimony. Meta CEO Mark Zuckerberg had been expected to testify but never took the stand.
Before trial, Meta said potential penalties could reach $1.4 trillion. The states suggested a figure closer to $200 billion. Neither amount was awarded because the case ended before a decision.
Meta does not admit wrongdoing
The judgment expressly states that it is a settlement only and “does not constitute an admission by Meta of any liability, wrongdoing, or violation” of law.
There was no verdict and no final judicial finding that the states’ allegations were true. Meta continues to deny wrongdoing.
“Ensuring teens have a safe and productive experience on our platforms is an absolute imperative for Meta,” the company said in its settlement announcement. Meta characterized the new terms as an extension of Teen Accounts and called on competing platforms to adopt comparable restrictions.
The agreement releases the participating governments’ covered consumer-protection and privacy claims. It does not settle personal lawsuits brought by individual users or cases filed by school districts and local governments. It also does not prevent New Jersey from enforcing other laws against future conduct outside the settlement’s releases.
The legal obligations apply in participating jurisdictions and cover Facebook, Instagram and qualifying future Meta social platforms. Existing products such as Threads, WhatsApp, Meta Quest and Meta AI are not automatically included.
Sherrill tied the settlement to three New Jersey laws she signed Aug. 11, including the New Jersey Kids Code Act, creation of a state Social Media Research Center and a study of possible warning labels for addictive social media features.