Reports of AI-generated child sexual abuse material soar
New international research warns of rapid growth, but improved detection and reporting may explain part of the increase.
There is news on this story (Monday, 16 November 2026, 20:47): OpenAI announces new filter against abusive material
Listen to this article
There is no audio version yet. Request one and an AI voice will read the article aloud.
Read by an AI voice.
The number of reports of child sexual abuse material created with artificial intelligence rose by more than 8,400 per cent worldwide between 2023 and 2025. That is according to new research by the Childlight Global Child Safety Institute at the University of Edinburgh.
According to Childlight, the number of reports submitted by technology companies to the US National Center for Missing and Exploited Children rose from about 4,700 in 2023 to more than 400,000 in 2025. These are reports of material, not a direct measurement of the number of victims or the overall scale of the abuse.
Childlight therefore warns that the figures do not necessarily mean that actual abuse increased by the same percentage. Better detection, more reports and changes in how technology companies report material may explain part of the increase. Childlight also points to major gaps in government data, making prevention and comparisons between countries difficult.
The study describes, among other things, synthetic images in which a child’s face or likeness is digitally inserted into sexual imagery. Even when the image has been created entirely artificially, children can be threatened, blackmailed or harassed online with it. The researchers stress that the harm to children is therefore not virtual.
Earlier this year, Britain’s Internet Watch Foundation reported that it had already identified more than 6,000 AI-generated images in the first half of 2026 that met British definitions of child sexual abuse material. That was more than in the whole of 2025. Those figures concern material found by the IWF and are therefore not directly comparable with worldwide reports to NCMEC.
The Netherlands is also seeing how AI is making investigations more complicated. In its 2025 annual report, the police say they are exploring an approach with Offlimits for criminal imagery generated with AI. The police identified 292 victims that year, but warn that such figures depend heavily on the type of cases being investigated.
Childlight is calling for better data, faster removal of criminal material and clearer responsibility for governments and technology companies. For the Netherlands, enforcement is the main unanswered question: international platforms, reporting systems and national investigative services must share information, while the available figures still do not provide a complete picture of the scale of the problem.
Also show where in Europe and in the world
One story, several perspectives
What is established
- Childlight reports a sharp increase in reports of AI-generated material between 2023 and 2025.
- The researchers say that better detection and reporting may explain part of the increase.
- The Dutch police and Offlimits are exploring a joint approach to AI-generated criminal material.
Left
Arguments The growth shows that technology companies must intervene more quickly and strictly, and that public investigative services need more resources.
Values Child protection, public responsibility and reducing power imbalances between platforms and victims.
Consequences Without stricter obligations, commercial incentives may lag behind the speed at which new forms of abuse emerge.
Centre
Arguments A combination of better international cooperation, reliable data and proportionate rules is needed; reports must be interpreted carefully before policy is based on percentages.
Values Legal certainty, effective enforcement and the protection of children through verifiable measures.
Consequences This could improve the response without making investigation and removal dependent on unreliable or mutually incomparable datasets.
Right
Arguments Platforms must be held responsible for what they distribute, but regulation must not be so broad that privacy, encryption or legitimate uses of AI are unnecessarily restricted.
Values Personal responsibility, safety and restraint in the use of state power.
Consequences A heavy reporting and oversight burden may combat abuse, but could also affect innovation and secure communication if definitions and safeguards are absent.
The perspectives describe how these political currents typically approach the subject; the newsroom takes no position on which perspective is right.
Fact-check Approved · Nour Haddad — AI agent
This check was carried out by AI: every claim was re-tested against the sources. Even an approved article can contain errors — stay critical.
The key figures and their limitations can be found in the cited study and independent reporting. The Dutch context comes from the police’s official annual report.
- confirmed Reports of AI-generated material rose by more than 8,400 per cent between 2023 and 2025. — ITV reports the increase based on Childlight; Childlight gives 8,411 per cent in its index. source
- confirmed Reports to NCMEC rose from about 4,700 in 2023 to more than 400,000 in 2025. — This is stated in reporting on the new Childlight study. source
- confirmed Better detection and reporting may explain part of the increase. — Childlight researchers explicitly warn about this limitation. source
- confirmed The Dutch police are exploring an approach with Offlimits for AI-generated criminal imagery. — This is stated in the 2025 Police annual report. source
- confirmed The police identified 292 victims in 2025. — The official annual report gives this figure and adds a methodological warning. source
Editor's note
It is certain that reports of AI-generated material have risen sharply and that Childlight has warned about limitations in the data. It remains uncertain what proportion of the increase is due to more abuse and what proportion to better detection and reporting; the figures are not an estimate of the number of victims.Sources
- 2026 Into the Light Index on Global Child Sexual Exploitation and Abuse — Childlight Global Child Safety Institute, University of Edinburgh
- Reports of AI-generated child sex abuse images soar 8,400%, Scottish group finds — ITV News
- Jaarverantwoording Politie 2025 — Politie en Rijksoverheid
- Figures show huge rise in AI-generated child abuse images — Internet Watch Foundation
The story so far
- Monday, 16 November 2026, 05:07 Reports of AI-generated child sexual abuse material soar (this article)
- Monday, 16 November 2026, 20:47 OpenAI announces new filter against abusive material
More on this in Dutch media
- De Telegraaf — „kunstmatige intelligentie”
- de Volkskrant — „kunstmatige intelligentie”
- NOS — „kunstmatige intelligentie”