Meta's $17.1 Billion Settlement Bans Filters and Like Counts. The Algorithm We Reported On Gets an Opt-Out, Not a Fix.

Meta is paying $17.1 billion to settle claims that its algorithm harms teenagers, and the settlement doesn't require Meta to change that algorithm's default behavior. The deal was approved by Judge Yvonne Gonzalez Rogers of the Northern District of California on Wednesday, August 26, 2026, cutting short a trial that had opened in Oakland eight days earlier. It requires Meta to pay at least $12.1 billion to the states over ten years, with roughly $5 billion more contingent on YouTube and TikTok agreeing to comparable safety measures — a structure confirmed independently by the attorneys general's offices of New York and the District of Columbia, and by CBS News's direct reporting on the order. The underlying lawsuit was originally filed in 2023 by a 29-state coalition led by California, Colorado, Kentucky, and New Jersey; Meta's own statement on the deal describes it as reached with 52 attorneys general across states, territories, and Washington, D.C., a wider figure reflecting states that joined after the original filing.

Photo by KATRIN BOLOVTSOVA on Pexels

This site asked a narrower question on August 30: an internal Meta study reviewed by Reuters found that teens who already felt bad about their bodies were served eating-disorder-adjacent content at roughly three times the rate of other teens — 10.5 percent of their feed versus 3.3 percent — and that Meta's own screening tools caught only 1.5 percent of what the company's own definitions would flag as sensitive. Our own piece, built on that reporting, asked what a recommendation system that over-serves in one direction might be doing to plus-size content on the other. The settlement resolves a related but different case. It is now the fullest documented answer to a narrower question: what did regulators and Meta actually agree needed fixing, and does any of it reach the mechanism our reporting described?

Photo by MART PRODUCTION on Pexels

Read directly against the attorneys general's own announcements — California's Rob Bonta, New York's Letitia James, and North Carolina's Jeff Jackson among them — the settlement requires Meta to cap daily use for users under 18 at 2 hours by default, block notifications between 10 p.m. and 7 a.m. and during the school day, restrict overnight access between midnight and 6 a.m., stop displaying like and reaction counts to minors, and ban cosmetic-procedure image filters — the kind marketed as simulating plastic surgery, including filters that change skin tone — for the same age group. It also requires stronger age verification, a faster harmful-content reporting system with a 90 percent response rate within 6 hours, and an independent auditor with ongoing access to Meta's data. Every one of those items is a real, enforceable change to a tool minors can use or a metric they can see. None of them is a change to what Meta's system decides to put in front of a teenager in the first place.

Photo by Sora Shimazaki on Pexels

One settlement term does name the algorithm directly. Both Bonta's office and James's office describe a requirement that minors be given an option for a "non-personalized feed" — chronological posts from accounts they already follow, instead of a ranked, recommended one. Jackson's office describes the same provision as letting teens select a feed "that isn't determined by Meta's algorithm." That is a real, verifiable term, and it is the only one in the settlement that touches recommendation logic at all. It is also, on its own wording, optional — a teen, or a parent, has to actively switch away from the default. Nothing in any attorneys general's announcement, and nothing in Meta's own statement, requires Meta to change what the algorithmic feed itself serves to teens who don't switch — the same default feed Reuters' reporting and our own piece described over-serving eating-disorder-adjacent content to teens who were already struggling. No settlement term requires Meta to improve the 1.5 percent detection rate its own research reported, either.

Meta's own statement on the settlement, attributed to chief legal officer C.J. Mahoney, frames the deal as an industry starting point rather than an admission — "because teens move fluidly across dozens of apps, we need an industrywide solution," the statement reads, calling on "our industry peers, TikTok and YouTube, to implement this new framework right away." It reads less like contrition than a dare — match what we just agreed to, or explain to your own regulators why you won't. Neither that statement nor any other Meta announcement found for this piece mentions body image, eating disorders, or algorithmic over-serving; it stays within general addiction and data-privacy framing throughout. Andy Stone, the Meta spokesperson who commented on the Reuters research for our August 30 piece, does not appear to have commented specifically on the settlement — no outlet found for this piece attributes a settlement statement to him by name.