Is Social Media Safe for Children?
Separating what is settled from what is genuinely contested.
Ask that question out loud and you will get two confident answers. One says social media is rewiring a generation and the only responsible move is to keep children off it entirely. The other says the panic is recycled from comic books and television, the research does not hold up, and the real risk is cutting kids off from where their friends are.
Both camps have credentialed people in them. Both have published. And a parent trying to decide whether a twelve-year-old gets an account this year is stuck between them with no good way to referee.
So here is an attempt at an honest answer, which means separating the parts that are settled from the parts that are not.
What is actually settled
A specific set of harms is measured, counted, and reported by organisations whose job is to count them. These are not contested.
Financial sextortion is the clearest case. The National Center for Missing & Exploited Children received more than 50,000 reports of financially motivated sextortion in 2025, up from more than 36,000 in 2024, which works out to an average of 137 reports a day (NCMEC, published 25 June 2026; accessed 7 November 2026). The pattern is consistent: someone poses as a peer, obtains an image, then demands money. It is fast, it is scripted, and it is run at volume.
The second settled harm is synthetic sexual imagery of real children. Thorn surveyed 1,200 people aged 13 to 20 and found that 1 in 8 personally knew someone who had been targeted by deepfake nudes, and 1 in 17 said images had been made of them (Thorn, "Deepfake Nudes & Young People," 3 March 2025; accessed 7 November 2026). On the supply side, the Internet Watch Foundation identified 3,443 AI-generated child sexual abuse videos in 2025, a 260-fold increase on the previous year, with 65 percent in the most severe category (IWF 2025 Annual Data & Insights Report; accessed 7 November 2026).
Nobody in the research debate disputes these numbers. They describe crimes committed by people against children, and they are increasing.
What is genuinely contested
The contested part is the big one: whether social media use is causing a population-level decline in adolescent mental health.
In December 2023 the National Academies of Sciences, Engineering, and Medicine published a consensus report on social media and adolescent health. Its conclusion was careful and worth quoting fairly: the committee's review of the literature did not support the conclusion that social media causes changes in adolescent health at the population level, and the evidence of association was not sufficient, on balance, to justify recommendations built on causal claims (National Academies, Social Media and Adolescent Health, December 2023; accessed 7 November 2026).
Candice Odgers made the same case in Nature in March 2024, reviewing Jonathan Haidt's The Anxious Generation. Her argument was that hundreds of researchers have looked for large effects and found a mix of null, small, and inconsistent associations, and that where longitudinal associations do appear, they are at least as consistent with young people who already have mental health difficulties using these platforms more and differently (Odgers, Nature, March 2024; accessed 7 November 2026).
Haidt disagrees, publicly and in detail, and he is not a crank. He argues the effect sizes are understated by how the studies measure exposure, and that the timing of the trend lines is hard to explain any other way.
A parent does not have to pick a winner. What matters is what the disagreement implies. The two sides are not arguing about whether a child can be harmed online. They are arguing about whether average use, by an average teenager, causes average harm. That is a different question from the one most parents are actually asking.
The policy experiment that is running right now
Australia gives us the closest thing to a test. Its ban on social media accounts for under-16s took effect on 10 December 2025. Platforms removed, deactivated, or restricted more than 4.7 million accounts assessed to belong to under-16s by mid-January 2026.
Three months in, the regulator's own findings were sobering. Nearly 70 percent of parents whose children had accounts before the ban said their children still used Facebook, Instagram, Snapchat, or TikTok. The eSafety Commission found no clear decline in complaints from under-16 users about cyberbullying or image-based abuse, while cautioning that three months is too short to conclude much (Australia eSafety Commission compliance findings, March 2026, as reported by Tech Policy Press; accessed 7 November 2026).
The honest reading is not that the ban failed. It is that a blunt instrument applied to a motivated teenager produces workarounds, and that removing accounts is not the same as removing risk.
A better question than "is it safe"
"Is social media safe for children" treats it as one substance with one dose. It is not. A thirteen-year-old watching cooking videos and a thirteen-year-old receiving unsolicited messages from adults are doing different things on the same app.
Three questions tell you more than the safety question does.
Who can start a conversation with my child? Almost every documented harm above begins with contact from a stranger. The sextortion scripts, the grooming, the image requests. A platform where unknown adults can open a conversation is a different product from one where they cannot, regardless of what else it does.
What is the product optimising for? If the feed is ranked to maximise time spent, that goal is not aligned with your child's evening. This is not a contested claim about mental health, it is a description of how the business works.
What happens to what my child posts? Who can copy it, who can download it, how long does it stay, and what is it used to train.
Those three are answerable per platform, this week, without waiting for the research to settle.
Where we come at this from
We build A2Me, a private family platform with no ads, no algorithmic feed, and no strangers. That makes us interested parties, and you should read us accordingly. It also means we have a specific and limited claim: removing strangers and ranking from one space your family uses does not make the rest of the internet safe. It just means that when your kid posts a photo of their birthday, the audience is their grandmother and not an open index.
The rest of the work is still yours, and most of it is conversation rather than configuration.
---