de-amplify_
← all distillations

distillation · the evidence ledger

State of New Mexico v. Meta: the standalone child-safety action

State of New Mexico v. Meta Platforms, First Judicial District Court, Santa Fe; New Mexico Unfair Practices Act; child-safety / exploitation theory

This is the raw, evidence-tiered claim ledger behind the New Mexico case file. It is published for transparency: the source of truth, with its tiers intact, current as of 2026-08-28.

how to read the tiers

Each claim is tagged by the strength of the evidence behind it, not by how certain it sounds.

  • [ESTABLISHED]: stated in a primary source (a court order or filing, the docket, an official release) or, for a hearing, said on the official record. A fact about what was filed or said, not necessarily adjudicated true.
  • [OBSERVED]: carried by secondary coverage, or by a document read into the record.
  • [ASSUMED]: the distiller's inference, flagged as such.
  • Provenance flags ([interested-party], [advocacy-witness], [single-witness], [lawmaker characterization]) mark a claim that rests on one interested voice: reported, not adjudicated.

Tier key (this record's calibration, per the parent plan): ESTABLISHED means stated in a primary source (a court order or filing, the docket, an official government release, including its verbatim republication) or in the project's verified baseline with independent corroboration; OBSERVED means carried by secondary coverage or inferred from it; ASSUMED means the drafter's inference, flagged as such. This authority-based calibration is the plan's deliberate adaptation for legal records, not the distillation skill's agreement-based multi-source tiers.

Reader summary

As of August 13, 2026, New Mexico is the first U.S. state to win a full trial against Meta (the company behind Facebook and Instagram) over harm to children, and the case has now been decided. On March 24, 2026, a Santa Fe jury found Meta broke New Mexico's consumer-protection law by misleading the public about how safe its apps were and by failing to protect kids from sexual exploitation, and ordered Meta to pay $375 million, the maximum the law allows ($5,000 for each of 75,000 violations). Meta says it will appeal, so that money is not final. A separate second phase, decided by Judge Bryan Biedscheid rather than a jury, ended in a final judgment file-stamped August 6, 2026 and announced August 7. The judge found Meta's platforms are a public nuisance in New Mexico, held that Section 230 did not shield Meta from that claim, and ordered it to pay a further $567 million into an abatement fund, bringing the total to $942 million.

The judge also ordered changes to how the apps work for minors, supervised for five years: no push notifications to under-18 accounts between 10:00 PM and 7:00 AM, or during school hours in term time; "like counts" hidden by default on under-18 accounts unless a parent permits otherwise; and a hard cap of 90 hours a month of use across Facebook and Instagram combined.

What the court refused is as important as what it ordered, and most coverage omits it. New Mexico asked for changes to the feed itself. The judge said no, and granted "no abatement relief relative to the design and implementation of Meta's algorithms." He agreed that autoplay, infinite scroll, like counts and algorithmic recommendations "combine to facilitate addictive or problematic behaviors in adolescent users," but held that ordering changes to them would directly affect how content is presented and so "would likely violate the First Amendment and Section 230", that such industry-wide features "require legislative or executive branch action", and that restricting Meta alone could unfairly damage it against competitors who were not in the case. The measures that survived (notifications, like counts, time limits) are the ones the court called "least connected with platform content." This is a child-safety and consumer-protection case; it is not the separate federal "social-media addiction" litigation.

Claims (the ledger)

(a) The case, court & theory

  1. [ESTABLISHED] The case is State of New Mexico v. Meta Platforms, Inc., a standalone action brought by the State of New Mexico. (Verified baseline; NM DOJ release.)
  2. [ESTABLISHED] It was tried in Santa Fe, in New Mexico's First Judicial District Court. (Verified baseline; corroborating coverage.)
  3. [ESTABLISHED] The presiding judge is Bryan Biedscheid of the First Judicial District Court. He was elected chief judge of that district in 2022, so both stylings in the record are correct: the DOJ's April 2026 release says "District Judge Bryan Biedscheid," and trial coverage says "Chief Judge." (Santa Fe New Mexican, 2022, on the chief-judge election; DOJ release via republication; Santa Fe New Mexican and Source NM trial coverage.)
  4. [ESTABLISHED] The legal claim in the first phase was brought under the New Mexico Unfair Practices Act (the state's consumer-protection statute). (Verified baseline; DOJ release; NPR; Crowell & Moring.)
  5. [ESTABLISHED] New Mexico Attorney General Raúl Torrez filed the suit in New Mexico state court in December 2023. (Crowell & Moring client alert; DOJ coverage.)
  6. [ESTABLISHED] The state's core theory was that Meta made misrepresentations about how safe its platforms were and failed to protect children from exploitation on those platforms. (Verified baseline; DOJ release; NPR adds the jury's related findings: false or misleading safety statements, unconscionable trade practices exploiting children's vulnerabilities, and concealment of knowledge of child sexual exploitation.)

(b) The $375M verdict + its statutory basis

  1. [ESTABLISHED] On March 24, 2026 (a Tuesday), a jury returned a verdict of $375 million in civil penalties against Meta. (Verified baseline; DOJ release; NPR; Source NM; Crowell & Moring.)
  2. [ESTABLISHED] This was a jury verdict (not a judge's ruling and not a settlement). (Verified baseline; NPR; Source NM.)
  3. [ESTABLISHED] The $375 million is the statutory maximum: $5,000 for each violation, applied to 75,000 violations ($5,000 × 75,000 = $375,000,000). (Verified baseline; DOJ release confirms "$5,000 per violation," "the maximum allowed," and "75,000 state law violations"; NPR confirms the $5,000 per-violation maximum.)
  4. [OBSERVED] The DOJ characterized the outcome as making New Mexico "the first state in the nation to prevail at trial against a major tech company for harming young people." (DOJ release, carried by search index; nmdoj.gov itself 403s, an access failure rather than a disconfirmation.)
  5. [OBSERVED] In the first phase New Mexico had sought a substantially larger sum: the state's closing argument urged a penalty that could top $2 billion (reported by CNBC as roughly $2.1 billion), and NPR notes the $375 million award was less than one-fifth of the request. (Source NM; NPR; CNBC via search. Enrichment only; the load-bearing figure is the $375M verdict.)

(c) What it was for: safety / exploitation, NOT addiction

  1. [ESTABLISHED] This is a consumer-protection / child-safety case about (1) misrepresentations regarding platform safety and (2) failure to protect children from sexual exploitation. (Verified baseline; DOJ release.)
  2. [ESTABLISHED] It is not an "addiction" case. (Verified baseline; flagged explicitly to prevent conflation.)
  3. [ESTABLISHED] It is not part of the federal multidistrict litigation (MDL). It is a separate New Mexico state-court action. (Verified baseline; Crowell & Moring confirms it was filed in New Mexico state court, a separate state action.)
  4. [OBSERVED] Trial evidence included internal Meta documents and testimony (from former Meta employees, law enforcement, and New Mexico educators) that the state said showed Meta's design features enabled predators to engage in child sexual exploitation on its platforms. (DOJ release / search summary; context only.)

(d) The appeal posture

  1. [ESTABLISHED] Meta has said it will appeal the $375 million verdict; the verdict is therefore not final / not affirmed as of the as-of date. Posture precision (corrected 2026-07-16): this is announced intent, not a filed appeal. The DOJ's own release says Meta "has announced its intent to appeal that verdict"; Meta's statement in verdict-day coverage is future-tense ("We respectfully disagree with the verdict and will appeal"); and no source reports a notice of appeal filed in this case, consistent with the second phase (and so any final judgment) still being open. An earlier version of this record said "Meta is appealing"; the deployed website content said the same and was corrected on 2026-07-16. A filed notice may appear at any time; that would be an update, not a reversal. (DOJ release via republication; NPR; Source NM.)
  2. [OBSERVED] Before the second phase, Meta also tried to dismiss the case, claimed immunity under Section 230, and moved to postpone the second-phase bench trial; the court denied those motions. (DOJ release via republication; context on procedural posture, not the appeal itself.)

(e) The second phase + the ~$953M abatement REQUEST + the design-change stakes

  1. [ESTABLISHED] There is a second, separate phase of the same case that is decided by a judge (a bench trial), not a jury. (Verified baseline; Santa Fe New Mexican; DOJ release.)
  2. [ESTABLISHED] The second phase concerns a public-nuisance claim and prospective (forward-looking) relief. (Verified baseline; DOJ release; Santa Fe New Mexican.)
  3. [ESTABLISHED] The second-phase bench trial began May 4, 2026. (DOJ release via republication: "Phase 2 begins May 4"; Santa Fe New Mexican.)
  4. [OBSERVED] The bench trial ran 13 days. (Santa Fe New Mexican.)
  5. [ESTABLISHED] Testimony in the second phase finished on May 21, 2026. (Verified baseline; Santa Fe New Mexican, re-fetched 2026-07-16: "testimony concluded May 21, 2026." May 22 was the phase's final day overall, not the end of testimony; that distinction was the source of an earlier page error, fixed 2026-07-16.)
  6. [OBSERVED] Both sides filed proposed findings / written closing statements around June 12, 2026. (Santa Fe New Mexican; search corroboration.)
  7. [ESTABLISHED] New Mexico has asked the judge to order Meta to pay approximately $953 million, which the state frames as Meta's equitable share (about 21%) of a $3.7 billion total 15-year abatement program; the ~21% is the portion an expert attributed to social media. (Verified baseline; Santa Fe New Mexican, fetched directly: "$953 million," Meta's "equitable share of the cost of abating," about 21% of a $3.7 billion total; Source NM: "nearly $1 billion.")
  8. [ESTABLISHED, and now SUPERSEDED by the judgment; see section (g)] That $953 million was a request, not an award, and the court did not grant it: the abatement fund ordered on 2026-08-06 is $567 million, not $953 million. The rest of this claim is retained as the pre-judgment record, including the caveat that ended up being the right one. The second-phase decision was still pending (undecided) as of the previous as-of date. Re-checked 2026-07-25: still pending. The New Mexico DOJ's live press-release index covers July 9 through July 24, 2026 and carries no ruling release, and this office has press-released every favourable milestone in this case; Meta's newsroom is likewise silent through July 24. (Verified baseline; Santa Fe New Mexican: "a request, not an award; the case awaits Judge Bryan Biedscheid's decision.") Caveat, stated because it is the honest limit of this check: absence of a press release is not the docket. The New Mexico Courts case-lookup application is gated behind a data-use agreement invoking a criminal-penalty statute, which was not accepted on the project's behalf. Anyone wanting certainty should accept that disclaimer themselves and search D-101-CV-2023-02838.

25a. [ESTABLISHED, correcting an earlier framing] The state's requested figure moved up, not down. New Mexico DOJ's own release of May 22, 2026 states the State sought $779.5 million; the ~$953 million figure appears on June 17, 2026. The $3.7 billion is an expert's 15-year program estimate introduced at opening, so describing $953 million as a reduction "down from $3.7 billion" adopts the State's own rhetorical framing and hides that the ask rose between May and June. State the sequence, not the spread. (NM DOJ release, 2026-05-22; Santa Fe New Mexican, 2026-06-17.)

25b. [ESTABLISHED, correcting a mischaracterisation] The NM DOJ release headlined "After $375 Million Loss, Meta Tries to Run. Court Says No." is dated April 9, 2026 and does not describe an appeal, a bond, or a stay. It describes Judge Biedscheid denying Meta's motion to postpone the phase-two bench trial. The same release says Meta "has announced its intent to appeal", which is announced intent only: no notice of appeal, supersedeas bond, or stay dispute has been found anywhere. (NM DOJ release 2026-04-09, recovered in full via the Internet Archive capture 20260506195252 because the live URL returns 403.) 26. [OBSERVED] The requested payment would fund a roughly 15-year program to abate the harm the state alleges ("abate public harm to children and teens"). (Santa Fe New Mexican; verified baseline.) 27. [OBSERVED] Beyond money, New Mexico asked the judge to order specific platform / design changes. The Santa Fe New Mexican's account of the state's request lists: "more effective age controls"; removing features such as "infinite scroll during sleep and school hours"; "safer algorithms" that "do not prioritize engagement over well being"; and mandatory advisories warning about "potential mental health harms." The Crowell & Moring alert describes the sought orders more broadly as including age verification, predator removal, and restrictions on encrypted messaging for minors; both accounts can be true of one broad request. (Santa Fe New Mexican; Crowell & Moring.) 28. [OBSERVED, and OVERTAKEN BY EVENTS; see section (g)] Why this case was watched: it was the live U.S. proceeding furthest along toward a court-ordered product-design change rather than only a payment; its remedies record was closed and awaiting decision. That decision landed on 2026-08-06 and this claim is now history rather than status. The design remedy exists, and it is narrower than the phrase "design-change order" suggests: notification, like-count and time-limit measures on under-18 accounts, and nothing at all on the feed. Corrected 2026-07-16: an earlier version called it "the only live proceeding in the American record that could order design changes," which is false as stated. The state attorneys general in federal MDL 3047 also seek injunctive relief (the October 2023 filing seeks "injunctive and monetary relief"; the June 29 order treats the request as live; the joint pretrial statement says injunctive relief "will be tried solely to the Court" in a remedies phase after liability), so the federal case could also reach a design-change order, but only in a post-liability remedies phase after its August 2026 trial. (Drafter synthesis over: the MDL primary filings via CourtListener/RECAP, the DOJ and SFNM accounts of this case's posture. See the audit issue for the full evidence.)

(f) Current status

  1. [ESTABLISHED] As of August 13, 2026: final judgment has been entered. The $375 million jury verdict stands, the court added a $567 million abatement fund and a five-year injunctive program, and Meta has said it will appeal (no filed notice found), so none of it is final. (The judgment itself; superseding the July 16 status this claim previously carried.)

(g) The final judgment, 2026-08-06 (read directly; this section is anchored to the order, not to coverage)

  1. [ESTABLISHED] The case was decided by a document captioned "FINDINGS OF FACT, CONCLUSIONS OF LAW, AND JUDGMENT, ORDER, AND DECREE OF THE COURT", 68 pages, file-stamped 8/6/2026 4:41 PM by the First Judicial District Court, Santa Fe County, and signed by Bryan Biedscheid, Chief District Court Judge, Division VI. This resolves the open question the Tensions section carried about the ruling's procedural label: it is all of those instruments in one document. Event date is 2026-08-06; the New Mexico DOJ announced it on 2026-08-07. (Judgment, caption and signature block.)
  2. [ESTABLISHED] The judgment recites the two-phase structure: Phase 1 a jury trial "held between February 2, 2026 and March 24, 2026" on the Unfair Practices Act claims and civil penalties; Phase 2 a bench trial "held between May 4, 2026 and May 22, 2026" on the State's abatement request based on public nuisance. (Judgment, opening paragraph.)
  3. [ESTABLISHED] The court ordered Meta to "pay and deposit a total of $567,000,000.00 into a fund", in addition to the jury's $375,000,000 Phase 1 civil penalties. Total exposure in New Mexico is therefore $942 million. The fund is allocated by category in a table in the order: Treatment $420,000,000.00; Screening and Assessment $90,000,000.00; Awareness and Prevention $33,000,000.00; Referral, Linkage, and Coordination $15,000,000.00; Implementation, Continuous Quality Improvement, and Evaluation $9,000,000.00. (Judgment ¶ 121 and the table following it.)
  4. [ESTABLISHED] The court held that Section 230 does not shield Meta from the public-nuisance claim: "This Court concludes that Meta is not immune from the public nuisance claim", and "the Court concludes that Section 230 does not protect Meta from the State's public nuisance claim." (Judgment ¶¶ 104 and the conclusion of section E.)
  5. [ESTABLISHED] The injunctive relief actually ordered on problematic use is three measures, all aimed at under-18 accounts. (a) Meta "shall eliminate push notifications on its platforms for known or estimated accounts belonging to users under 18 years of age" from 10:00 PM to 7:00 AM on all days and from 8:00 AM to 3:00 PM on academic-year weekdays, with exclusions permitted for messaging from connected users and for "urgent, targeted messages, such as for security or hazard alerts." (b) Meta "shall hide, as a default setting for all accounts where the user is under 18 years of age", all like counts it appends to content, overridable only with parent or guardian permission. (c) Meta "shall implement a mandatory usage time limit" restricting under-18 users to "not more than 90 hours of use per month cumulatively across Facebook and Instagram." (Judgment ¶¶ 168 to 170.)
  6. [ESTABLISHED, and the load-bearing claim for this project] The court granted no relief at all against the feed: "The Court grants no abatement relief relative to the design and implementation of Meta's algorithms." Its reasoning is explicitly a First Amendment and Section 230 reasoning, and it is the mirror image of claim 33: Section 230 did not bar liability, but it did constrain the remedy. The state's algorithm requests were "vague and aspirational, rather than objective and measurable", "would directly alter and limit the manner in which platform content is displayed", and so "would likely violate the First Amendment and Section 230." (Judgment ¶¶ 171 to 173.)
  7. [ESTABLISHED] On infinite scroll and autoplay specifically, the court declined on separation-of-powers and competitive grounds: "regulation of industry-wide features that contribute to problematic use, such as infinite scroll and autoplay, require legislative or executive branch action", and ordering changes "would risk competitively damaging Meta's platforms to an unreasonable extent, and possibly running afoul of the First Amendment and Section 230 because of the direct effect those features have on content presentation." The court also noted "the absence of Meta's competitors in this litigation". (Judgment ¶¶ 160, 164.)
  8. [ESTABLISHED] The court nonetheless made an explicit finding on the mechanism, which is the part of this order most useful to a design-standard argument: "the evidence at trial proved that design elements, such as autoplay, infinite scroll, ... and push notifications create a product that, particularly for adolescent users, is highly rewarding psychologically and neurochemically. As a result, for many, it is irresistibly compelling to start scrolling and very difficult to stop or abstain from its use." It further found that "algorithmic content recommendations can create harmful feedback loops and" what it called "rabbit hole" users. (Judgment ¶¶ 161 to 162.)
  9. [ESTABLISHED] The line the court drew between what it would and would not order is a content-proximity line, stated in those terms: push notifications and like counts are "least connected with platform content", a like count is "merely a feature created and offered by Meta to principally track and motivate user feedback, all without altering the underlying published content", whereas autoplay, infinite scroll and algorithmic recommendations "are most closely tied to content presentation because they directly impact the manner in which users are presented with third-party postings and advertising." (Judgment ¶¶ 163, 165.)
  10. [ESTABLISHED] On the problematic-use measures the court was "in large measure, adopting Meta's proposals", finding "Meta's proposed measures to address push notifications on Facebook and Instagram to be compelling." (Judgment ¶ 160.) [ASSUMED] that this materially weakens any reading of these provisions as a contested win on design; the court says it adopted the defendant's own offer. (Drafter inference from ¶ 160, flagged.)
  11. [ESTABLISHED] The order runs for an Abatement Period with reporting requirements, and creates a Child Safety Monitor; it also contains sections on Child Sexual Exploitation, Chatbot Usage, Accurate Disclosure and Educational Programs, and Law Enforcement Reporting. (Judgment, section headings B, D, E, H, I, J of the abatement part.) Those sections have not been distilled claim by claim. Age Assurance (section C) has been, below, because it bears directly on the policy paper's section 7.

(g.1) Age Assurance: the court wanted it, and held that COPPA forbids ordering it

  1. [ESTABLISHED] The court treated age verification as central rather than peripheral: "Age verification is the key to making Meta's platforms safe for adolescents", because there must be substantial certainty about whether a user is over 13 (old enough to use the platforms) and over 18 (so adolescent protections come off and restrictions on connecting with adolescents go on). Without that certainty, it found, there will be "adults communicating with adolescents by claiming a false younger age and tweens communicating with teens and adults by claiming a false older age". (Judgment ¶ 136.)
  2. [ESTABLISHED, and the finding this project should carry into section 7] COPPA is what blocked the remedy, not the First Amendment. The court reasoned that the COPPA Rule defines "collection" broadly and expressly includes "passive tracking of a child online", so ordering age verification would compel the very collection COPPA restricts. Its conclusion: "it is the Court's conclusion that it cannot order Meta to request children to submit personal data or be passively tracked online, even for age-verification purposes." The section is captioned accordingly: "The Children's Online Privacy Protection Act limits this Court's options to require age-verification tools." (Judgment ¶¶ 134 to 135 and the section C.i heading.)
  3. [ESTABLISHED] The court considered the FTC's 2026 age-verification enforcement policy and declined to rely on it, because the policy "does not create any substantive rights or entitlements", the FTC "retains its ability to bring actions for violations of COPPA through age-verification tools", and agencies "are free to change their existing policies" so long as the change is explained (citing Encino Motorcars, LLC v. Navarro, 579 U.S. 211 (2016)). Flatly: "The FTC policy from 2026 does not change the COPPA Rule." (Judgment ¶¶ 134 to 135.) [OBSERVED] The policy referred to is an FTC statement announcing it "will not bring an enforcement action under the Children's Online Privacy Protection Rule (COPPA Rule) against certain website and online service operators that collect, use, and disclose personal information for the sole purpose of determining a user's age via age verification technologies", limited to general-audience and mixed-audience services. (ftc.gov press release, read 2026-08-13.)

43a. [ESTABLISHED, and the reason the site names that policy by year only] The statement carries no date at all, which is why no month can honestly be attributed to it. The policy statement itself was retrieved on 2026-08-13 and cached at docs/distillations/sources/ftc-2026-02-25-coppa-age-verification-policy-statement.txt. It bears no date anywhere on its face: no issue line, no "By direction of the Commission", no vote line, no signature block. A search of the full three-page text for any month-day-year string returns nothing. That is the simplest explanation for the judgment's own vagueness at ¶ 134, where a court with the document in front of it wrote only that the FTC "issued a policy in early 2026". It could not cite a month either.

 **The surrounding date evidence is nonetheless clear, and is recorded here rather than on the site.** Three independent signals give **February 25, 2026**: the statement PDF's own `CreationDate` (Wed Feb 25 10:50:29 2026), the date printed under the press-release headline, and that page's `article:published_time` metadata. One outlier, a bare `<time>` element on the release page reading February 13, 2026, is supported by nothing else and remains unexplained. So the release is confidently dated and the statement is undated, which is why the paper names it as the FTC 2026 enforcement-discretion policy and stops there. An earlier draft of that sentence gave it a month on the strength of law-firm summaries; the month is probably right, and it was still the wrong way to get it.

 **Retrieval note, corrected 2026-08-14.** `ftc.gov` returns HTTP 403 to a default `curl` User-Agent and **HTTP 200 to a browser User-Agent**, verified five times in a row on the policy-statement PDF including with the exact header combination that had failed earlier. An earlier draft of this claim said the block was TLS-fingerprint based and that a browser User-Agent did not defeat it. **That was wrong.** It generalised from a single 403 that was probably transient rate-limiting, and it is the same error this record warns about elsewhere: a confident mechanism inferred from one under-tested probe. The three government and court domains this project fetches do NOT behave alike, which is why the blanket version was wrong: `ftc.gov` gates on User-Agent, `nmdoj.gov` returns 403 to `curl` whatever the User-Agent, and `courtlistener.com` does not block `curl` at all (its 403s are specific to the sandboxed fetcher, not to the domain). Chrome's PDF viewer separately ignores injected scroll and keyboard input, so a PDF opened there still has to be downloaded to be read in full.

44. [ESTABLISHED] The same COPPA problem defeats the machine-learning route. The court found that Meta's "efforts to train a classifier to predict the age of users in this group is hindered by COPPA and its prohibition on the use of data that is needed to train a classifier to be more accurate in its age predictions", adding that it was "not criticizing the policies behind COPPA" but noting its effect. (Judgment ¶ 139.) 45. [ESTABLISHED] The court split the problem by age band. Users near 18 "likely have" signal (internet activity, a driver's license, passport, consumer history) sufficient to determine or prove age "with relatively little effort on the part of the user". Users near 13 "likely do not have identification, credit, numerous transactions in their name, or other" such signal. (Judgment ¶¶ 137 to 138.) 46. [ESTABLISHED] Beyond COPPA, the court gave three institutional reasons, each of which maps onto an existing limit in the policy paper. (a) Separation of powers: "the regulatory solutions to these problems lie in the executive and legislative branches and not with this Court", reinforced by the observation that the UK, EU and Australia limits "were all promulgated by their respective parliaments and not by courts in individual cases". (b) Single defendant: "creating significant friction at sign up for only Meta would be inequitable and unduly injurious to Meta". (c) Displacement: such barriers would "merely shift users to other, non-party platforms". (Judgment ¶¶ 138 and 140.) 47. [ESTABLISHED] The court also found the most promising technical routes simply outside its reach: solutions "such as phone service providers or app stores sharing information with Meta, are outside this Court's jurisdiction", and it repeated that cellphone-company and app-store verification is "unavailable to this Court for a lack of jurisdiction over any such parties". (Judgment ¶¶ 138 and 141.) 48. [ESTABLISHED] What it ordered instead was Meta's own proposal, and it said so without enthusiasm: the court concluded that "the following abatement relief proposed by Meta, while far from ideal or sufficient, is most appropriate". It rejected the State's proposals, including requiring the youngest users "to acquire identification or appear before a judge to obtain a judicial determination of their age", as "unreasonable and would result in an unworkable barrier to the use of Meta's platforms by those who have a right to do so". (Judgment ¶¶ 140 and 142.)

(g.2) Age Assurance: what the court actually ordered

The reasoning is at (g.1). This is the operative list, and it is the most concrete specification of a child-safety brake any American court has entered against a platform.

  1. [ESTABLISHED] The order requires Meta to keep improving its age-assurance models and tools in New Mexico by two named routes: using "Artificial Intelligence tools to enhance Meta's heuristics, including by incorporating additional signals", and attempting to develop, within two years, "a dedicated under-13-years-of-age prediction model". The qualifier is part of the order and must travel with it: that attempt is on reasonable best efforts "in light of COPPA limitations", which is claim 44's problem written into the remedy. Source: the judgment, paragraph 142(a). Note for re-verification: the words attempting to develop, within two years straddle a page break in the cached text, so they are stated here rather than quoted. They are deliberately left unquoted, because quoting them is exactly what check:quotes rejects.

  2. [ESTABLISHED] Where Meta's systems predict a New Mexico account holder is under 13, Meta must proactively request proof of age and mark for deletion the accounts of those who do not verify within 30 days, subject to legal-hold obligations. It must also proactively assess profiles, images, posts and comments for an admitted age inconsistent with the stated age. Source: paragraph 142(b) and 142(c).

  3. [ESTABLISHED, and this is the most transferable single measure in the order] Where Meta believes a user is under 13, or under 18 but cannot estimate a precise age, it must treat the user as being in that band unless and until the user verifies. That is a default-to-protective rule on uncertainty, and it is exactly the shape of remedy this project argues for: it does not require the platform to know a user's age, only to behave a particular way when it does not. Source: paragraph 142(d).

  4. [ESTABLISHED] Three enforcement-side measures. On an age-related suspension or deletion of an under-13 account, Meta must review the friend network to assess whether any friends are also under 13. Item (f) of the order reads "Partner with schools or a child safety organization to create a reporting portal where school administrators can flag suspected accounts", and that portal covers suspected under-13 accounts "across any social media platform", not only Meta's. And reporting must be made easier "by reducing the number of steps and information required". Source: paragraph 142(e), (f), (g).

  5. [ESTABLISHED] Meta must delete the accounts and all personal information collected from under-13 accounts, and must maintain "an age-assurance appeal process" for users "who contend their age was estimated incorrectly". Source: paragraph 142(h) and (i).

  6. [ESTABLISHED, and it is the reason this section matters to the argument] Every measure at claims 49 to 53 is Meta's own proposal, adopted because better ones were unavailable to the court. Read with claim 48, where the court calls that relief "far from ideal or sufficient", and with claim 47, where the most promising technical routes are outside its jurisdiction, the order says something a refusal alone does not: what a court could reach here, it knew was not enough. The refusal at claims 36 to 38 shows a court declining to touch the feed. This shows the same court granting what it could and recording its inadequacy on the face of the judgment. Source: paragraph 142 read with paragraphs 140 to 141.

Tensions / open questions

  • A settlement New Mexico is not part of will generate headlines that sound like it is (opened 2026-08-28). On 2026-08-26 Meta settled with 51 states and territories in the federal MDL for up to about $17.2 billion. New Mexico is in none of it: it has no payment row in that agreement's Exhibit B, it is not among the 48 "Cambridge Settling States", and this case, already at final judgment, was never in that litigation. The agreement does enumerate New Mexico's other Meta case, the 2021 Cambridge Analytica matter (State of New Mexico ex rel. Torrez v. Facebook, Inc., No. D-101-CV-2021-00132), as one of four template complaints, and then does not release it, because New Mexico is not a releasing state. So both New Mexico cases stand entirely apart from the settlement: this one at judgment with an appeal expected, the Cambridge one still pending. Coverage that folds "the states settled with Meta" over New Mexico is wrong twice. Source: the settlement agreement, MDL Dkt 572-1, Exhibits B and E and Sections I.L and I.O, cached in sources/ and verified in the mdl-3047 ledger, claims 46 and 60.

  • The $3.7 billion figure is RESOLVED (2026-07-16). A May 13, 2026 U.S. News headline referred to Meta "Challeng[ing] New Mexico's $3.7 Billion Plan for Teen Mental Health." A direct re-fetch of the Santa Fe New Mexican reconciles it cleanly: $3.7 billion is the total 15-year cost an expert estimated to abate New Mexico's youth mental-health crisis "as a whole," and ~$953 million is Meta's equitable share, about 21% (the portion another expert attributed to social media). The two figures are not in conflict: $953M is Meta's slice of the $3.7B total. Adopted.

  • "Addictive" language does appear, but in the nuisance context, not as an addiction case. The state's second-phase public-nuisance theory refers to "easy access to addictive social media platforms" alongside "exploitation at the hands of sexual predators" (Santa Fe New Mexican). This wording is about the harm the nuisance claim targets; it does not convert the case into the separate federal "social-media addiction" litigation. The case's core (the phase-1 UPA verdict) is consumer-protection / child-safety. Kept distinct deliberately.

  • Exact procedural label of the pending decision. RESOLVED 2026-08-13 by reading the order. It issued as a single document doing all of it at once, captioned "FINDINGS OF FACT, CONCLUSIONS OF LAW, AND JUDGMENT, ORDER, AND DECREE OF THE COURT". See claim 30.

  • The judgment cuts both ways on Section 230, and the site must not report only the favourable half. Section 230 did not bar liability on the public-nuisance claim (claim 33), and that is the half the Attorney General's release leads with. The same order then relied on Section 230 and the First Amendment to refuse remedy against the algorithm, infinite scroll and autoplay (claims 35 and 36). For a project whose thesis is that the regulable surface is the delivery loop, the second half is the more consequential finding of the two, and it arrived from a judge who had just accepted the mechanism as proved. Any use of this case on this site that mentions the win without the refusal is a misrepresentation by omission.

  • This is one trial court, and it is not final. A New Mexico district court's constitutional reasoning binds nobody, Meta has said it will appeal, and the algorithm holding is the kind most likely to be contested. Do not state the First Amendment / Section 230 conclusions as settled law; state them as what this court held.

  • The problematic-use measures were largely Meta's own proposals (claim 39), which makes them weaker evidence of what a court will impose over objection than the headline suggests.

  • Age Assurance remedies distilled 2026-08-18, and the note that prompted it was overstated. A dedupe pass over the MDL trial coverage reported that this ledger did not carry four claims about this order's age-assurance remedies, and said the paragraph 142 finding should be added with them. The reasoning was already here, at claims 41 to 48, and the paragraph 142 finding was already claim 48. What was genuinely missing was the enumerated remedy list itself, now distilled at section (g.2). Recorded because the note is a worked example of the failure it was written about: a claim about what a repository lacks needs the same reading pass as a claim about what a source says.

  • Sections of the order not yet distilled. Child Sexual Exploitation, Chatbot Usage, Accurate Disclosure, Law Enforcement Reporting, and the Child Safety Monitor are summarized only at heading level (claim 40). Age Assurance came off this list on 2026-08-18, distilled at (g.1) and (g.2); until then this bullet named it first and called it the one that matters most to the paper's section 7, which is now served. (For a day this bullet and the one above asserted opposite things about the same section, which is the cost of updating a ledger by appending: a correction that lands as a new bullet has to also revisit the bullet it corrects.)

  • Appeal watch. A notice of appeal may be filed once the case reaches an appealable judgment (likely after the phase-2 decision). When that happens, claim 16 updates from "has said it will appeal" to "has appealed"; the as-of dating is the guard.

Sources

← the New Mexico case file

all evidence ledgers