The $942 Million Question: Did Social Media Help Create a Youth Mental Health Crisis?

Of the new $567 million fund, $420 million is earmarked specifically for treatment services for young people, with the remainder directed to awareness and prevention campaigns, screening and assessment programs, and referral coordination over the next five years. Judge Biedscheid’s language was notably blunt. He compared Meta to a polluting factory, writing that “although Meta is not alone in this regard, its social media platforms are a significant contributing factor to the current mental health crisis among New Mexico’s youth established by the substantial evidence in this case.”

Beyond the money, the order mandates a set of operational changes:

The judge rejected some of the state’s more sweeping requests, including changes to Meta’s core algorithms and features like infinite scroll and autoplay, after Meta argued during trial that some requested changes were not technologically feasible and could force it to exit the state entirely.

Meta has said it disagrees with the ruling and will appeal, maintaining that it has been transparent about the difficulty of removing bad actors and stands by its record on teen safety.

The most visible proponent of a strong causal claim is NYU social psychologist Jonathan Haidt, whose 2024 book The Anxious Generation: How the Great Rewiring of Childhood Is Causing an Epidemic of Mental Illness argued that the shift from a “play-based childhood” to a “phone-based childhood” beginning around 2010 to 2012 is a primary driver of a documented rise in adolescent anxiety, depression, and self-harm in the U.S. and UK. Haidt, together with researchers Zach Rausch and Jean Twenge, maintains two continuously updated “collaborative review” documents that compile published studies on adolescent mood disorders since 2010 and on social media’s relationship to mental health, arguing the pattern amounts to more than correlation. Their recommended policy response, no smartphones before 14 and no social media before 16, has been influential in shaping legislative proposals and school phone bans across multiple U.S. states.

This position draws support from the U.S. Surgeon General’s office. In 2021, then-Surgeon General Vivek Murthy issued an advisory on protecting youth mental health, and in a June 2024 New York Times op-ed, he called on Congress to require a tobacco-style warning label on social media platforms, stating social media is associated with significant mental health harms for adolescents. Murthy was explicit that he could not impose such a label himself, since it would require congressional legislation, and cited research finding that adolescents who spend more than three hours a day on social media face roughly double the risk of anxiety and depression symptoms.

Longitudinal research forms part of the evidentiary basis for this view. A frequently cited 2019 study published in JAMA Psychiatry, which prospectively followed 6,595 U.S. adolescents, found associations between time spent on social media and later internalizing problems, and combined internalizing and externalizing problems, after controlling for baseline mental health status measured before the social media exposure, a methodological choice aimed at addressing the “reverse causality” objection that struggling teens simply migrate to social media rather than social media causing the struggle. Notably, the study’s adjusted association with externalizing problems alone did not reach statistical significance, a nuance often lost when the study is cited as blanket support for social media’s harms.

The scientific consensus is considerably less settled than the headline version of Haidt’s argument suggests, and serious researchers have pushed back hard on both his data and his causal framing. A 2024 review in Science, authored by the journal’s editor-in-chief, characterized the underlying science as “unsettled,” noting that Haidt’s ability to capture the public zeitgeist has outpaced the strength of the experimental evidence. Psychologist Candice Odgers, reviewing The Anxious Generation in Nature, went further, arguing the book overstates causality that the underlying studies do not support, a critique that has itself been contested by Haidt’s collaborators as selectively engaging with the evidence base.

The core scientific objection is about effect size and study design. Critics point out that much of the large body of studies Haidt cites is observational and correlational rather than experimental, and that where randomized experimental designs do exist, the measured effects of social media on wellbeing tend to be small, comparable in magnitude to other mundane behavioral correlates that nobody would call a public health crisis. Some researchers have also raised the possibility that part of the apparent rise in diagnosed adolescent mental illness reflects changes in diagnostic criteria, screening practices, and healthcare access over the same period, rather than a pure behavioral effect of screens.

The American Psychological Association’s own 2023 Health Advisory on Social Media Use in Adolescence, arguably the most institutionally authoritative document on this question prior to the Surgeon General’s 2024 escalation, takes a deliberately calibrated position. Its lead finding states plainly that using social media is not inherently beneficial or harmful to young people, and that effects depend heavily on the individual adolescent’s pre-existing psychological characteristics, social context, and what, specifically, they are doing online: passive scrolling versus active social connection, exposure to harmful content versus access to supportive communities. The advisory explicitly flags that causal data is hard to obtain partly because the richest behavioral data sits inside the technology companies themselves and has not been made available to independent researchers, a structural limitation that cuts against confident claims in either direction. The APA also notes a genuine protective effect for some vulnerable groups: it cites evidence that social media access provides meaningful psychological benefit to LGBTQIA+ and questioning adolescents, who may find peer support online that is not available to them offline.

These are not, in the end, fully irreconcilable positions, even though they are frequently staged as if they were. The Haidt camp’s strongest claims rest on population-level trend correlation combined with a plausible mechanism: engagement-optimized design, social comparison, sleep displacement, and the “like” economy. The skeptical camp’s strongest claims rest on the point that individual-level experimental effect sizes are small, and that a headline correlation at the population level does not establish that the mechanism is dominant, causal, and uniform across a heterogeneous population of tens of millions of teenagers.

Both of those things can be true simultaneously. Social media can be a real, non-trivial contributing factor for a meaningful subset of adolescents, particularly those already vulnerable, and particularly around specific design features like infinite scroll, algorithmic amplification, and appearance-based metrics like “likes,” without being the single dominant explanatory variable for a complex, multi-causal mental health trend that also implicates academic pressure, reduced unstructured play, sleep deprivation, economic precarity, and a documented post-2020 rise in youth mental health strain tied partly to the pandemic.

This is, notably, close to where Judge Biedscheid’s own order lands, legally rather than scientifically. He did not find that Meta’s platforms are the sole or even primary cause of New Mexico’s youth mental health crisis, but that they are “a significant contributing cause,” a standard for public nuisance liability, not a claim of exclusive causation. That is a more careful formulation than either “social media is destroying a generation” or “the panic is overblown,” and it happens to track the actual state of the peer-reviewed literature reasonably well.

The dollar figures will dominate coverage, but the more analytically interesting part of the ruling is the list of mandated product changes, because they target the mechanisms that the more careful research literature, as opposed to the more polarized public debate, actually implicates.

The hidden “like” count provision is directly responsive to a mechanism Meta’s own internal research reportedly identified years earlier: Instagram’s effect on teen girls’ body image, reported publicly via the 2021 Facebook Files disclosures. In other words, the order is less a verdict on whether “social media causes depression” as a monolithic claim, and more a mechanism-by-mechanism intervention on the specific product features that policymakers and researchers have identified as plausible mechanisms of harm, which is arguably a more defensible legal and policy posture than the sweeping causal claims found in either the popular advocacy literature or the platforms’ own defenses. Notably, the judge declined to go further and mandate changes to core engagement mechanics like infinite scroll or autoplay, on both technical-feasibility and evidentiary grounds, a reminder that even this order does not treat all proposed mechanisms as equally well-established.

Two things remain genuinely open, and neither is resolved by this ruling.

This outcome is a data point in a much larger, still-unfolding legal landscape rather than a final word, and Meta’s appeal means even this judgment is not yet settled as precedent.