SANTA FE, 11 AUG 2026 — A New Mexico judge has ordered Meta to pay US$567 million into a fund for treating the harm its platforms did to children in the state, and has told the company to change how Facebook and Instagram behave for minors there. He did not touch the recommendation algorithm, and he explained why.
The reason is the same statute we wrote about last week when the Ninth Circuit held that Section 230 is a defence rather than a bar to being sued. Here, the statute moves from liability to remedy, shaping what the court felt able to make Meta do.
The ruling
First Judicial District Judge Bryan Biedscheid issued a 68-page decision on Thursday in Santa Fe. He found Meta's platforms a public nuisance under New Mexico law and a "significant contributing cause" of a youth mental health crisis in the state.
He located the harm in the business model, not in any particular piece of content. His formulation of the nuisance is precise: "Meta's platforms create a public nuisance because their purpose and effect is to optimize engagement, including in ways that are detrimental to teenagers' health and safety."
The $567 million goes into an abatement fund, of which $420 million is for treating young people already harmed. The remaining $147 million covers awareness and prevention, screening, and related costs over five years. A jury had already awarded the state $375 million in civil penalties in March, bringing what Meta has been ordered to pay in New Mexico alone to $942 million.
What the injunction reaches, and what it does not
[sentence deleted]
| Ordered | Declined | |
|---|---|---|
| Product | Time limits for underage users in New Mexico | Changes to the recommendation algorithms |
| Defaults | Hide "like" counts on minors' photos by default | Ending end-to-end encryption on Facebook |
| Disclosure | Visible risk warnings shown to those users | — |
| Reporting | Easier reporting of underage use, plus a portal letting school administrators flag suspected under-13 accounts across any platform | — |
On the algorithms, CNBC reports that Biedscheid wrote ordering changes could conflict with Section 230 and First Amendment protections. He also declined to require Meta to stop supporting end-to-end encryption on Facebook.
Read those two exclusions together and a boundary appears. The court was willing to reach the interface — what a teenager sees, how long they can stay, what the defaults are. It was not willing to reach the ranking system that decides which posts appear, or the cryptography that decides who can read a message.
Why that boundary matters more than the money
$942 million is a real number and a smaller one than it sounds for a company of Meta's size. The durable part of this ruling is the doctrine and the remedy, because both are portable.
Public nuisance is an old tort about interference with a right common to everyone, and it was the theory that produced the opioid settlements after product liability proved awkward. Applying it to a social platform sidesteps the question that has swallowed most of these cases, which is whether the complaint is really about third-party content. A nuisance claim can be framed around the design of the system rather than the posts it carries, and design is not what Section 230 was written to protect.
But the remedy shows the limits of that framing. If the harm is engagement optimisation, and the engine that performs engagement optimisation is the recommendation algorithm, then a remedy that excludes the algorithm is treating the symptom the plaintiff itself identified as the mechanism. Biedscheid ordered Meta to warn users about the product and to limit how long they can use it, while leaving intact the thing that decides what it shows them.
What the case was built on
Attorney General Raúl Torrez brought it after an undercover investigation in which staff created accounts posing as children. He told CNBC that a fabricated profile of a 13-year-old girl "was simply inundated with images and targeted solicitations."
The method is doctrinally significant. Evidence generated by the state's own accounts describes what the system does to a new minor user, rather than what any particular adult did, which is the kind of proof a design-defect argument needs and a content-moderation argument tends to lack.
Torrez framed the outcome as transferable: "For years, Meta knew its platforms were harming New Mexico's kids... and it chose engagement and profit over their safety." Meta says it will appeal, and that it works hard to keep people safe and has been transparent about the difficulty of identifying bad actors.
Why this reaches beyond New Mexico
At least 40 state attorneys general have filed similar suits, and a federal case brought by a large group of states is pending in California. New Mexico has now produced a decided outcome with a written remedy, which gives every one of those plaintiffs a template and every defendant a preview.
The read for anyone building a consumer product with minors on it is narrower than the headline. The orders that survived here are about defaults, session limits, disclosure and reporting routes — all things a product team controls directly and can change without touching ranking. That is a fairly precise description of what a court in the United States currently believes it can require, and it is a reasonable design brief regardless of jurisdiction.
What to watch
The appeal, and specifically whether the public nuisance theory survives it. A reversal on the doctrine would take the template away from the other 40 states.
Whether Meta applies the ordered changes only to New Mexico. State-by-state product variation is expensive and awkward, and companies facing one such order have often generalised it — which would make a single state's remedy the de facto national default.
And whether any court is willing to go where this one would not. Every meaningful case against a recommender system eventually arrives at the same question. So far, the answer has been that the interface is reachable and the ranking is not.