Meta and a bipartisan coalition of state attorneys general announced a proposed settlement Wednesday morning, ending the trial that began last week in Judge Yvonne Gonzalez Rogers’ Oakland courtroom.
The agreement calls for payments of up to $17.1 billion over 10 years and requires Meta to change how Facebook and Instagram operate for users under 18. It remains subject to court approval.
The settlement applies to the attorneys general’s claims. It does not resolve the personal-injury, school-district or municipal cases pending in the federal MDL, the California JCCP or other courts. Those claims are expressly preserved in the proposed consent judgment.
Instagram chief Adam Mosseri began testifying Tuesday afternoon and was due back on the stand Wednesday. Mark Zuckerberg was expected to appear later in the trial. The settlement takes both executives off the witness schedule in the states’ case.
What Meta Agreed to Change
The agreement imposes default restrictions on Facebook and Instagram accounts belonging to users under 18. Most of the requirements will remain in effect for 10 years.
The principal terms include:
- A two-hour cumulative daily limit across Facebook and Instagram. A parent must approve additional time.
- A default block on platform access between midnight and 6 a.m.
- Push notifications muted between 10 p.m. and 7 a.m. and during school hours.
- Usage prompts after 15 minutes of continuous activity and again after 60 and 90 minutes of cumulative daily use.
- An option for teenagers to use a chronological feed instead of Meta’s personalized recommendation system. Parents using Meta’s supervision tools may require that setting.
- Like and reaction counts hidden from users under 18.
- Restrictions on cosmetic-surgery and extreme-makeup filters.
- Stronger systems for identifying users under 18 and removing accounts operated by children under 13.
- An improved process for reporting harmful content. Meta must respond to 90% of those reports within six hours.
- Additional parental alerts involving secondary accounts and potentially suspicious contacts.
An independent auditor will monitor compliance and issue public summaries of its findings. The agreement also prohibits Meta from making false or misleading statements about the effectiveness of its safety features.
Direct messaging is generally excluded from the daily and overnight limits.
The settlement contains additional provisions tied to TikTok and YouTube. If those companies agree to comparable restrictions and financial terms, Meta’s daily limit would drop to one hour per platform and its nighttime block would expand to 10 p.m. through 7 a.m. Part of Meta’s maximum payment is contingent on that industry participation.
Meta made no admission of liability or wrongdoing. The settlement itself cannot be offered as evidence in other proceedings except in matters involving its approval or enforcement.
The Testimony That Preceded the Settlement
The agreement arrived after two days of testimony about optional safety features, default settings and the information reaching Instagram’s senior management.
On Monday, San Diego State University psychology professor Jean Twenge spent nearly four hours on the stand as an expert for the states. J. Christian Lewis of the Kentucky Attorney General’s Office handled the examination. Twenge reviewed research documenting increases in adolescent depression, self-harm and suicide beginning around 2010. Girls between the ages of 10 and 17 showed some of the largest increases.
She testified that extended social-media use can displace sleep, exercise and in-person relationships, particularly when use exceeds approximately two hours per day. “That is a terrible formula for mental health,” Twenge told the jury.
Meta attorney Paul Schmidt challenged the studies because many did not separate Instagram and Facebook from other social-media platforms, video games or general screen use. He also emphasized the difference between evidence that social media may contribute to poor mental health and proof that Instagram caused a particular injury. The afternoon testimony moved from the scientific literature to Instagram’s own safety work.
California Deputy Attorney General Megan O’Neill called George Volichenko, a former Meta data scientist who worked on Instagram’s mental well-being team during 2022 and 2023.
Volichenko worked on “Take a Break” and “Quiet Mode.” He testified that few teenagers meaningfully used the features. Under one measurement discussed in court, approximately two of every 1,000 weekly teen users completed a prompted break.
Employees considered making the protections automatic instead of requiring teenagers to activate them. Volichenko said management resisted that approach because of the anticipated effect on engagement. “The tradeoff to core metrics was not desirable,” he testified.
Volichenko left Meta after approximately nine months. He said his team had limited freedom to develop and launch changes that could produce meaningful results. He also testified that the team served, in part, as protection against anticipated litigation.
Meta attorney Brian Stekloff established on cross-examination that Volichenko is a data scientist, not a clinician or adolescent mental-health expert. Volichenko agreed that adolescent mental health is complicated and that qualified employees at Meta took safety seriously.
Internal Data Removed from a Leadership Presentation
On Tuesday, Colorado attorney Jason Slothouber questioned Instagram product design director Francesco Fogu about a 2023 presentation prepared for company leadership. An earlier version reportedly stated that teenagers were exposed to approximately 1.5 times more content involving bullying, suicide, hate, nudity and violence than adults. That information did not appear in the final presentation.
Fogu acknowledged that the material had been removed but rejected the suggestion that it was taken out to conceal unfavorable information from Instagram’s leadership. Slothouber also displayed an internal passage describing how the company could respond to regulation: “Sometimes we don’t comply and accept a fine. Sometimes we comply in the most minimal or literal way possible.”
Fogu repeatedly said he could not remember whether he wrote it. Judge Rogers eventually stopped the repeated questioning, telling counsel, “We get the point.” Fogu also acknowledged that Instagram understood fewer teenagers would use “Take a Break” if the feature was not enabled automatically. That testimony preceded an agreement requiring Meta to place teenage users into daily time limits, overnight blocks and notification restrictions by default.
Mosseri’s Testimony Ends After One Hour
Mosseri took the stand late Tuesday and testified for approximately an hour before court adjourned. He defended Instagram’s safety program and objected to measuring the company’s work through the performance of a single feature. He said Instagram has introduced numerous protections and continues to improve them.
Slothouber questioned him about the low adoption of optional safeguards and Instagram’s later introduction of broader automatic protections through Teen Accounts in 2024.
Mosseri denied encouraging employees to withhold negative information. “I can’t think of a single time when I encouraged people to give me less information,” he testified. He also acknowledged that access to sensitive teen research became more restricted after the 2021 Facebook Files disclosures. Meta said the restrictions protected sensitive information and ensured that research was handled properly. The states were examining whether the policy also limited the circulation of unfavorable findings inside the company. That examination will not be completed in this trial.
The Private Cases Continue
The proposed consent judgment specifically excludes claims brought by private individuals, school districts, municipalities and other plaintiffs with cases pending in the federal MDL, the California coordinated proceedings or other state and federal courts. No settlement funds announced Wednesday are designated as compensation for individual personal-injury claimants.
The agreement also does not decide whether Meta caused an individual plaintiff’s injuries. Meta preserved its defenses in the remaining litigation, including its challenges to general and specific causation. The testimony heard in Oakland nevertheless identified several areas that will remain important in the private cases: the adoption rates for Meta’s safety tools, the feasibility of automatic protections, the effect of those protections on engagement and what information reached senior management.
Any attempt to use the settlement itself will face Rule 408 and the consent judgment’s express evidentiary restrictions. The internal documents and testimony presented before the settlement raise separate questions that will be addressed through discovery and the normal evidentiary process.
The attorneys general obtained substantial payments and mandatory product changes before Mosseri completed his testimony and before Zuckerberg took the stand. The private cases now move forward with that abbreviated trial record available for lawyers to study.