Policymakers are moving fast on teen social media. The evidence suggests slower, smarter choices.
Governments from Canberra to Brussels are flirting with blunt tools: age bans, curfews and feature restrictions aimed at shielding teenagers from online harm. Those moves are driven in part by high‑profile arguments that smartphones and social media have “rewired” childhood. But leading developmental researcher Candice Odgers, who has studied adolescent mental health for 25 years and, since 2008, has led intensive digital‑tracking work with young teens, warns that headline remedies risk doing more harm than good.
What the policy rush looks like
Australia introduced a law barring under‑16s from social media that took effect in December 2025. The rollout required platforms to identify and disable accounts for Australian users under 16 and attached fines that could reach tens of millions of Australian dollars, according to reporting in the New York Times. The UK government’s expert panel has recommended measures such as a midnight to 6am curfew for older teens, and the EU has pledged stronger protections for under‑13s. Those moves reflect a political appetite for clear fixes, simple to explain but not always effective.
Two public stories about the same problem
The debate has two dominant narratives. One, popularized by Jonathan Haidt in The Anxious Generation, frames smartphones and social media as major causal drivers of rising adolescent anxiety and depression. Haidt’s book sold widely and has shaped public and political urgency; he has argued in public forums that parents should consider excluding their children from these platforms.
The other, advanced by researchers such as Odgers, stresses nuance. Odgers has said: “Hundreds of researchers, myself included, have searched for the kind of large effects suggested by Haidt. Our efforts have produced a mix of no, small and mixed associations.” In a TED talk she added bluntly: “Scary stories sell; they always have. And scary stories are really easy to sell to parents. We’re an anxious lot.”
Why blanket bans are tempting, and why they can backfire
Bans and curfews are politically attractive because they’re simple to explain: limit access, reduce exposure, protect children. But simplicity trades off with enforcement, unintended consequences, and opportunity costs.
- Enforcement is hard. The Australia rollout showed practical limits: some accounts were logged out, others remained accessible, and circumvention is straightforward where devices are shared or VPNs are available (New York Times reporting, December 2025).
- Forbidden spaces push risk underground. Odgers warns that forcing teens off mainstream platforms can drive them to encrypted apps, private group chats or foreign services where harmful behaviour is less visible and reporting chains are weaker. That “forbidden space” effect can reduce detection and intervention.
- Policy attention can be diverted. Large political energy spent on age bans may crowd out investment in school counselors, community supports and criminal enforcement that target the most severe harms.
The evidence: not clean, not absent, and not uniform
Neither “social media is the sole cause” nor “social media is harmless” captures the evidence base. Multiple expert reviews and panels emphasize complexity. The National Academies of Sciences, Engineering, and Medicine concluded that “contrary to the current cultural narrative that social media is universally harmful to adolescents, the reality is more complicated.”
What researchers consistently find is heterogeneity. Programs such as the Oxford‑based BrainWaves initiative report that longer time on social media correlates with stronger associations with depression and anxiety, but they also show that motives matter: recreational and social uses can be neutral or positive, while what BrainWaves calls “harmful motivation”, using platforms for revenge, compulsive upward comparison, or to seek validation after exclusion, tends to predict worse outcomes.
Methodological realities shape headlines. Most large studies are observational, so they document correlations rather than ironclad causation. Confounders, such as economic shocks, family mental health, COVID‑era disruption and substance exposure, and reverse causality, where struggling teens use social media differently, complicate interpretation. Where effects are reported, they are often small or concentrated in subgroups rather than uniform population‑level shifts. Odgers summarizes the field bluntly: researchers have “produced a mix of no, small and mixed associations.”
Serious online harms demand criminal and regulatory action
Not all online risks are subtle or small. Sextortion and image‑based sexual abuse are criminal acts with severe consequences for victims, and Odgers has urged prosecution of perpetrators and stronger platform responsibility for allowing these harms to occur. She has noted the unequal burden of these abuses: a large share of victims are women and girls, while the design and executive ranks of many tech firms remain heavily male, a point she raises as part of a broader accountability argument.
“The identification of social media as the culprit is misleading, it doesn’t show up that way in the vast majority of studies, ” Odgers has said.
Practical policy and design alternatives that are actionable
Instead of universal bans, Odgers and many child‑protection experts recommend a layered response: criminal enforcement, platform accountability, targeted supports for vulnerable teens, and carefully evaluated policy experiments. Below are steps leaders can take that are specific and measurable.
- Prosecute and interrupt criminal behaviour. Prioritize law‑enforcement capacity to investigate sextortion and image‑based abuse, and require platforms to preserve evidence and cooperate with investigations.
- Mandate transparency and enforcement metrics. Require platforms to publish quarterly enforcement dashboards showing removal rates for image‑based abuse, average time to action, number of reports received, and appeal outcomes. Those metrics make platform obligations auditable.
- Invest in human supports. Fund more school counselors, mental‑health services, and safe after‑school spaces so problems can be caught early. These are direct supports for kids that do not depend on perfect tech enforcement.
- Design child‑sensitive defaults. Push for age‑appropriate product design: friction on sharing intimate images, default privacy settings for minors, and limits on discovery algorithms that steer vulnerable teens toward harmful content.
- Target interventions, don’t punish everyone. Use screening tools in schools and clinics to identify teens using platforms with “harmful motivation” (revenge posting, compulsive comparison, socially reactive use) and deliver tailored supports instead of imposing universal access restrictions.
- Run policy experiments and rigorous evaluations. Roll out major changes in phases and embed evaluation. Use randomized or quasi‑experimental designs, such as phased rollouts across districts and difference‑in‑differences comparisons across regions, and measure outcomes like platform substitution, reporting rates, sleep, school attendance and clinical referrals.
How to read the Odgers and Haidt disagreement
Haidt frames the problem as urgent and large‑scale; Odgers acknowledges serious harms but disputes that the bulk of evidence supports a single‑cause, population‑level effect of social media on adolescent mental health. Their public exchanges, including a discussion hosted by the University of Virginia in October 2024, matter because policy is already shifting. When books sell in the millions and headlines crystallize fear, policymakers may act faster than evidence accumulates.
That matters practically. The Australia law (December 2025) is already a natural experiment: initial reporting showed enforcement gaps and anecdotal circumvention, and early follow‑up research has suggested many under‑16s continued using social media after the ban. A BMJ study reported that more than 85% of participating under‑16s in Australia said they were still using social media three months after the ban. That finding raises questions about enforcement, substitution, and reporting, and it needs careful peer review and replication to determine longer‑term effects.
Questions leaders ask, and concise, evidence‑grounded answers
-
Do social‑media bans for teens stop harmful use?
Early policy experiments show bans are difficult to enforce and many young people continue to access platforms or migrate to less‑regulated spaces, reducing visibility and reporting of harm rather than reliably eliminating risky behaviour (New York Times reporting on Australia; follow‑up studies such as the BMJ analysis noted above).
-
Is social media the main cause of rising teen anxiety and depression?
The bulk of large reviews and many research teams report mixed, small, or no large causal effects; social media is one factor among several (economic shocks, the pandemic, family mental health, substance exposure), and impacts vary by context, content and motivation for use.
-
Are all teens equally at risk online?
No. Risk is concentrated in subgroups: gender, preexisting mental‑health status, social context and use motives change vulnerability. Sexual exploitation and image‑based abuse disproportionately affect girls, and these harms require criminal justice responses.
-
What policy moves are most likely to reduce serious harms?
Investing in prosecution of perpetrators, enforcing platform responsibilities (transparency, preservation and rapid removal), and funding school and community mental‑health supports are more likely to reduce severe harm than blanket access bans that may simply move abuse into hidden spaces.
-
Should companies be held accountable?
Yes. Child‑protection advocates and many researchers call for tighter regulation: mandated reporting of harms, safety‑by‑design requirements, and legal consequences for platforms that fail to take reasonable steps to prevent known abuses.
A short decision checklist for leaders
- Do not substitute symbolic bans for operational solutions. Avoid unilateral, nationwide age bans without an evaluation plan and without commitments to fund enforcement and supports.
- Prioritize criminal enforcement and platform accountability in the near term. Ensure police and prosecutors have resources; require platforms to keep evidence, report metrics and design for safety.
- Launch pilots with rigorous evaluation within six months. Use phased rollouts or randomized designs, collect pre/post data and publish the results to inform scale‑up decisions.
Where to watch next
Watch the Australia rollout for longer‑term evaluation results and substitution patterns; monitor UK and EU policy instruments for how they define enforcement and platform obligations; and look for peer‑reviewed studies that exploit these policy changes as natural experiments. Researchers need longitudinal, causal designs to move beyond correlation and to identify which interventions reduce harm for which groups.
If you or someone you know needs help
UK: Papyrus 0800 068 4141; Samaritans 116 123. US: 988 Suicide & Crisis Lifeline at 988. Australia: Lifeline 13 11 14. For other international helplines, visit befrienders.org.
Policy urgency is understandable, the harms are real and sometimes devastating. But fast, headline‑friendly fixes that don’t match enforcement realities or the contours of the evidence risk making things worse. The safer route is targeted criminal enforcement, platform transparency and design obligations, investment in human supports, and disciplined policy experiments that tell us what works before we reshape childhoods at scale.