UNICEF study finds 60% of online child sexual abuse happens on social media

Around 20 million children experienced sexual exploitation or abuse online in a single year across 21 countries, according to a UNICEF study published on Thursday. That is roughly one in five internet-using children aged 12 to 17 in the countries surveyed.

The findings come from “Through Children’s Eyes: How Digital Technologies Enable Child Sexual Abuse”, produced under the Disrupting Harm project that UNICEF runs with ECPAT International and INTERPOL, and were reported by Reuters. UNICEF calls it one of the largest studies of its kind.

The methodology is what gives the number weight. Researchers ran nationally representative surveys of about 21,000 internet-using children across Africa, Asia, Latin America and Eastern Europe between 2020 and 2025, alongside 100 in-depth interviews with survivors.

Breaking the estimate down, more than 15 million children were exposed to unwanted sexual material, more than nine million were pressured into sexual conversations or into sharing images, and four million had intimate images shared without their consent. The categories overlap, since a child can experience more than one.

Almost 60% of the incidents happened on social media, with Facebook, Instagram, Snapchat, TikTok and WhatsApp named in the study. Another 14% took place in online games.

That concentration matters for the policy argument. The platforms carrying most of this harm are the same ones now facing age verification mandates in Europe and under-16 bans in several countries, and the study gives regulators a prevalence figure to point at rather than an anecdote.

Generative tools have opened a second front. Across nine of the countries surveyed, an estimated 1.1 million children said sexual imagery of them had been produced with AI, a category that did not meaningfully exist when the first round of fieldwork began in 2020.

That figure lands alongside warnings already being issued to parents about AI-generated abuse imagery. It also complicates detection, because systems built to match known material struggle with images that have never existed before.

The nine countries behind that figure match the second round of fieldwork exactly, run in 2024 and 2025. The twelve countries surveyed four years earlier were not asked, which leaves no baseline against which to measure how fast the problem is growing.

The perpetrator data cuts against the stranger-danger framing that shapes most parental advice. In 57% of cases, the person involved was already known to the child, a peer, a partner, a friend, or a family member, while 38% of cases began with online-only contacts operating fake accounts.

The consequences are severe and largely invisible. Children who experienced this abuse reported rates of suicidal thoughts and self-harm around four times higher than their peers, along with elevated anxiety.

Almost none of it reaches anyone who could act. Fewer than 1% of cases were reported to police or to any support service, which means the official statistics that shape enforcement priorities capture a rounding error of the actual scale.

UNICEF’s recommendation points at product design rather than parental vigilance. “Platforms must build safety into products from the outset, rather than relying on children and parents to manage risks after harm occurs,” said John Livingstone, head of digital policy at UNICEF Australia.

The survey covered two rounds. Cambodia, Ethiopia, Indonesia, Kenya, Malaysia, Mozambique, Namibia, the Philippines, Tanzania, Thailand, Uganda and Vietnam were studied in 2020 and 2021, with Armenia, Brazil, Colombia, the Dominican Republic, Mexico, Montenegro, North Macedonia, Pakistan and Serbia added in 2024 and 2025.

The absence of the United States, the United Kingdom and the EU member states from the sample is a real limit on what the estimate covers. It is a floor for 21 countries rather than a global total, and UNICEF has not extrapolated beyond them.

The findings arrive as regulators are already testing whether platforms are doing enough, with Australia’s eSafety commissioner finding serious gaps in how large companies respond to child sexual abuse and sexual extortion. UNICEF has separately found that children are adopting AI tools roughly three times faster than adults.