The number arrives every year with the same blunt weight, and every year it climbs. In 2025, the National Center for Missing & Exploited Children's CyberTipline logged 21.3 million reports of suspected child sexual exploitation, containing 61.8 million images and videos. That is not a rounding error in a policy brief. It is a scale of harm that did not exist a generation ago, generated almost entirely by infrastructure adults built, sold, and profited from, and are only now scrambling to fence in.
Online child safety has quietly become the defining tech-policy fight of the decade, cutting across privacy law, constitutional speech protections, artificial intelligence, and basic parenting.
What makes 2026 different is not that the problem is new. It's that the evidence base has finally caught up to the scale of it, and lawmakers on four continents are now legislating against numbers they can no longer wave away.
01. What the data actually shows
Start with bullying, the least sensational and most common harm. The Cyberbullying Research Center's 2025 national survey of 3,466 US middle and high schoolers found that 58% had experienced cyberbullying at some point in their lives — the highest figure the center has recorded since it began tracking in 2007, with a third of respondents targeted in just the past month.
The most common forms weren't dramatic; they were exclusion from group chats, cruel comments, and mockery of physical appearance, the digital-era update of the same cruelty children have always inflicted on each other, now available around the clock and impossible to leave at the school gate.
A more unsettling shift shows up underneath the headline number. For the first time since tracking began, boys now report higher past-30-day victimization than girls, 36.6% versus 28.6%, a reversal researchers attribute largely to the explosion in financial sextortion schemes targeting teenage boys.
Girls still face disproportionate rumor-spreading and image-based abuse, but the sextortion economy has broadened its target list, and the data is only starting to reflect it.
By the numbers — United States, 2025–26
Offline maltreatment data offers a sobering backdrop. Federal figures from the Administration for Children and Families show more than 532,000 confirmed child abuse and neglect victims in the most recent reporting year, with neglect accounting for nearly 80% of cases. Online and offline harm are not separate categories in a child's life; they compound. A neglected or unsupervised child is also, overwhelmingly, an unsupervised device user.
02. Where the harm concentrates
Platform-level research increasingly identifies where risk clusters rather than treating "the internet" as one undifferentiated space. Pew Research Center's 2026 platform analysis found that roughly three in ten Snapchat users had been called offensive names, had rumors spread about them, or been physically threatened on the platform, a figure that reframes disappearing-message apps, long marketed as low-stakes and casual, as sites of some of the most concentrated peer harassment measured.
"Legislation is not yet delivering the outcomes families need. Nearly half of children say the content they've seen online recently is more child-friendly, and nearly half still say age checks are easy to beat."
Internet Matters, Digital Wellbeing Index, 2026Britain offers the clearest early test case of what regulation actually changes, and what it doesn't. Since the UK's Online Safety Act took effect, 68% of children and 67% of parents report seeing more safety features, and 53% of children say they were recently asked to verify their age. That's a real behavioral shift. But 46% of children think age checks are easy to bypass, and 32% say they have actually done it, drawing on a fake birthday or, in one parent's account, an eyebrow pencil moustache that satisfied a facial-age scan.
Nearly half of children, 49%, say they still experienced some form of online harm in the past month. The law changed the interface. It has not yet changed the underlying vulnerability.
03. The global regulatory wave
What's notable about 2026 is not any single law but the sheer number of governments moving at once, and how differently they're moving. Australia went furthest first: its ban on social media accounts for under-16s took effect December 10, 2025, the first hard age floor of its kind anywhere. Singapore followed with an Online Safety Act targeting "image-based child abuse," explicitly including AI-altered and AI-generated material.
Vietnam has required parental registration of under-16 accounts since 2024 and tightened identity-check mandates further this year. Austria agreed to a under-14 social media ban in March 2026, with Belgium now weighing similar restrictions after a government health council report on youth wellbeing.
TAKE IT DOWN Act
Signed into law in May 2025, it creates a federal takedown mechanism for non-consensual intimate imagery, including AI-generated deepfakes, and is a current FTC enforcement priority for 2026.
KOSA & COPPA 2.0
The Kids Online Safety Act remains pending in Congress as of mid-2026. COPPA 2.0, reintroduced in 2025, would raise privacy protections to under-17s and bar targeted ads to minors.
Digital Services Act guidance
The European Commission's 2026 minors-protection guidelines push platforms toward a proportionate, risk-based approach to age assurance rather than a single mandated method.
Child Digital Safety Law
Requires platforms to run age-verification systems scaled to their risk level; full compliance is mandated by January 1, 2027.
In the US, the picture is more fragmented. States including New York and California have pushed baseline design-safety rules and risk-based audits, while federal proposals stall over exactly the question that keeps derailing this issue everywhere: who is responsible for verifying a child is a child, the platform, the app store, or the device manufacturer, and how much of a stranger's identity a teenager should have to hand over just to open an account.
04. Why this is not just a parenting problem
The instinct to treat this as a household issue, better parental controls, more screen-time limits, stricter family rules, is understandable and largely beside the point. UNICEF has long argued that the internet did not invent child exploitation, but it enhanced the scale and reach of abuse that already existed, giving offenders new means of contact and distribution.
Design choices made in Silicon Valley product meetings, autoplay, disappearing messages, algorithmic amplification, frictionless account creation, shape risk exposure far more than any individual parent's vigilance can offset.
Ninety percent of the burden currently falls on families anyway; UK researchers note that even after the Online Safety Act, much of the responsibility for managing online risk still falls on parents rather than platforms, which is precisely the imbalance the current wave of legislation is trying, unevenly, to correct.
Critics of bans and mandatory age verification aren't wrong that the tools are blunt. Civil liberties groups warn that hard age floors push younger users toward less regulated corners of the internet rather than off it entirely, and that identity-verification systems can chill free expression for teenagers who use the internet precisely because home isn't safe for them to be themselves.
Both things can be true at once: the current internet exposes children to real, measurable, escalating harm, and the fixes on the table are still crude instruments being tested in public, on real children, in real time.
What the 2026 data makes hardest to dispute is the direction of the trend line. Reports to NCMEC keep climbing. Sextortion keeps finding new victims. Age checks keep getting beaten with an eyebrow pencil.
Governments that spent a decade treating child online safety as a public-awareness campaign are now, visibly, treating it as an infrastructure problem, one that has to be solved in the code and the law, not just in the family living room. Whether the current round of legislation actually closes the gap, or simply moves it somewhere less visible, is the story the next year of data will have to answer.
.jpg)
.jpg)
0 Comments