The Brief
Society & Ethics 5 min read

8,400% in Two Years: AI Is Reshaping Child Exploitation

NAVION

Share

Reports of artificial intelligence being used to sexually exploit children have climbed by more than 8,400% in two years, according to research from Childlight, the Edinburgh-based Global Child Safety Institute. The institute counts AI-related child abuse material, including deepfakes made with “nudify apps”, rising from about 4,700 reports in 2023 to more than 400,000 in 2025. Behind that jump sits a harder question than any single statistic can answer: what does it mean when a general-purpose technology becomes a tool for harming children at scale?

What the numbers measure, and what they don’t

The Childlight analysis draws on reports of material filed by technology companies with the US-based National Center for Missing and Exploited Children. That detail matters. These figures are reports of material, not a direct count of victims or of individual crimes. A single child can appear in many images, and a single image can generate many reports.

Professor Debi Fry, Childlight’s global director of data, was careful on this point. She acknowledged that improved detection and reporting may explain some of the increase. But she described the rise as extraordinary and the speed of change as deeply concerning. The honest reading is that two things are probably happening at once: companies are getting better at finding this material, and the material itself is multiplying.

The images are fake, yet the harm is not. Childlight notes they can be used to blackmail children, whether to extract money or to demand further images. A fabricated picture still works as a weapon because the child fears it will be believed.

A second watchdog, the same direction

Childlight is not the only body seeing this pattern. The Internet Watch Foundation (IWF), a UK-based NGO that monitors abuse material globally, reported last month that AI-generated child sexual abuse material is growing online. By its account, the amount of illegal material investigated this year has already passed the total for all of 2025.

The IWF’s analysts found more photorealistic child sexual abuse material in the first half of 2026 than in the whole of the previous year. They assessed 6,310 AI images that met the legal definition of child sexual abuse, which is 40% higher than last year’s total of more than 4,500. The IWF also noted the problem had already risen in 2025.

Two organisations, using different data and different methods, point the same way. That convergence is what makes the trend hard to dismiss as a measurement quirk.

The wider Childlight research puts this in context. It estimates that more than 300 million children worldwide are victims of at least one form of technology-facilitated sexual exploitation or abuse every year, and that about one in four children will experience at least one form of abuse before turning 18. In North America, almost a third of girls and more than one in 10 boys were subjected to in-person sexual abuse during childhood, a category that includes sexual assault, rape and “upskirting”. Online and offline harm are not separate problems running on separate tracks.

Why the framing matters more than the headline

Most coverage of AI and abuse treats it as a story about a frightening new technology. The more useful lens is the one Childlight itself proposes: a preventable public health emergency.

That framing shifts the question. A technology story invites debate about what the tools can do. A public health story asks who is at risk, where harm concentrates, which interventions work, and whether governments are measuring any of it well. Paul Stanfield, Childlight’s chief executive, argued that governments too often work with an incomplete picture of where harm is happening and who is most at risk. His point is that you cannot prevent what you cannot see.

This is also where the data work comes in. Fry described combining evidence from surveys, police records and other sources to reveal weaknesses in protection systems and to target resources where they are needed most. The research covers east Asia and the Pacific, Latin America and the Caribbean, and North America, building on earlier studies of western Europe. Even the national comparisons hint at uneven exposure: the recorded rate of reports and notices involving child sexual abuse material rose by 63% in the US between 2024 and 2025, compared with 18% in Canada. The source does not explain that gap, which is itself a sign of how much remains unmeasured.

The phrase “preventable” carries weight. Stanfield stressed that there is nothing inevitable about child sexual exploitation and abuse, and that the knowledge and tools to make a difference already exist. The call is for sustained leadership, international collaboration, better evidence, and the conversion of that evidence into early intervention.

What this means for how we think about technology

The uncomfortable lesson is that capability and misuse arrive together. Image-generation tools built for ordinary purposes can be turned to exploitation, and “nudify apps” show how a narrow, harmful use can be packaged into something easy to reach. The speed here is the story: a roughly 85-fold increase in reports over two years is far faster than laws, platform policies or public understanding typically move.

For ordinary readers, the takeaway is not to retreat from technology but to demand that its risks be measured and managed with the same seriousness as its benefits. That means technology companies detecting and removing material quickly, as Fry urged, and governments investing in evidence rather than reacting after the fact. It also means recognising that human protection systems, from police to child safety specialists, need data and resources to keep pace with the volume AI can generate.

In short

Reports of AI-related child sexual abuse material rose from about 4,700 in 2023 to more than 400,000 in 2025, an increase of more than 8,400%, with the Internet Watch Foundation independently finding the problem still accelerating in 2026. Better detection may explain part of the rise, but both research bodies treat it as a genuine and fast-growing threat, one that can also be used to blackmail children. Childlight argues that governments should treat this as a preventable public health emergency, investing in better evidence and turning it into early action. The central point is that the harm is not inevitable: with sustained leadership, international cooperation and tools to measure the problem, it can be reduced.

Based on reporting from The Guardian.

Written by

NAVION