The Childlight Global Child Safety Institute in Edinburgh published a study on November 16, 2026 on how AI is being used to sexually exploit children. Reports topped 400,000 in 2025, up from roughly 4,700 in 2023. That's an 8,400% jump — roughly an 85-fold increase. The count is based on reports that tech companies file with the US National Center for Missing & Exploited Children (NCMEC).
The scheme is simple and vile. Deepfake services — the report calls them "nudify apps" — take an image of a child and produce a pornographic result. The image is fake, but it gets used as if it were real: to blackmail the child, demand money, or extort new photos. Deb Fry, Childlight's data director, calls the scale of exploitation "shocking." Institute CEO Paul Stanfield says governments often operate with an incomplete picture: it's unclear where the harm is actually happening and who is at risk.
For the first half of 2026, the Internet Watch Foundation assessed 6,310 AI images that meet the legal definition of such material. That's 40% more than in all of 2025, when more than 4,500 were counted. Reports and takedown notices for such material in the US rose 63% from 2024 to 2025; in Canada, 18%. Childlight estimates that more than 300 million children encounter at least one form of technology-facilitated sexual exploitation every year.
Part of the growth is detection growth — Fry admits as much herself. Better detection and more diligent reporting explain some share of it. But they can't account for the entire leap from 4,700 to 400,000. Our read: image generation got cheap. You don't need a big model — a small one with a web interface and a subscription costing a couple of dollars will do.
Reporting to NCMEC is, in effect, an American pipeline. It only contains what companies decide to send. A child in a country without such a channel may get no response at all. And one more thing: a fake image costs nothing to make, but the victim's fear is real. The "it's a fake, so it's not scary" logic breaks down here.
By our count, only two groups come out ahead: the operators of "nudify" apps and whoever hosts their servers where nobody asks questions. On the losing side: children and moderators. The IWF and Childlight get their numbers and their attention, but their reports ship faster than parliaments write rules.
The report includes no methodology. Are the 400,000 unique items, reports, or multiple notices about the same file? Which apps and models are involved isn't stated; how many cases made it to court — also unknown. The numbers come from the companies themselves, so they depend on the quality of each company's detection. The source is a single one: Childlight's publication covered by the Guardian; the IWF figures within it are secondhand.
Forecast: Within six months, at least one major jurisdiction — the EU or the UK — will write dedicated requirements for "nudify" services into law. Watch the bill texts, not the press releases.
Source: The Guardian
Source: www.theguardian.com · post in Telegram