A groundbreaking study released in February 2026 by UNICEF, INTERPOL, and the ECPAT global network delivered one of the most alarming statistics yet in the AI deepfake crisis: at least 1.2 million youngsters across 11 countries disclosed having their images manipulated into sexually explicit deepfake content in the past year. The study represents the most comprehensive attempt to quantify the scale of AI-generated child sexual abuse material (CSAM).

Why It Matters

This study transforms the deepfake debate from a hypothetical risk into a documented, quantified crisis affecting over a million children. It provides the empirical ammunition that legislators, regulators, and platform operators need to justify aggressive intervention — and shifts the conversation from "should we regulate" to "how fast can we act." For the sex tech and adult content industry, it reinforces the need for robust age verification and consent verification at every level of the AI content pipeline.

"Deepfake abuse is abuse," UNICEF warned in the report's release, calling on tech platforms and governments to treat AI-generated exploitation with the same severity as traditional CSAM production and distribution. The research highlights that generative AI tools have dramatically lowered the technical barrier to creating realistic fake intimate imagery, putting children at risk even from peers with no specialized technical knowledge.

The findings emerge against a backdrop of escalating regulatory action. The U.S. federal TAKE IT DOWN Act requires platform compliance by May 19, 2026. California's AB 621 deepfake pornography law took effect in January 2026. The EU AI Act's deepfake disclosure provisions are entering phased enforcement. Multiple states have introduced or passed bills targeting both individual creators and AI platforms that enable deepfake production.

The study's scope — spanning countries across multiple continents — suggests this is a global phenomenon that transcends any single platform or jurisdiction. Experts note that the 1.2 million figure likely underrepresents the true scale, as many victims may be unaware their images have been manipulated.

Sources

Update — 2026-03-14

Initial entry — story first created.

Update — 2026-03-31

The Internet Watch Foundation's 2026 annual report, published March 24, delivered staggering corroboration of the UNICEF findings: AI-generated child sexual abuse videos surged from just 13 in 2024 to 3,443 in 2025 — a 26,385% increase. Over 65% of the AI-generated CSAM depicted the most severe forms of abuse, with girls comprising 97% of victims. The IWF warned that deepfakes of specific children can now be created in as few as 15 minutes using only 20 source images. The report reinforces the urgency of the EU's nudification app ban (passed plenary March 26), the TAKE IT DOWN Act's May 19 compliance deadline, and the growing wave of state-level deepfake legislation.

New Sources

Update — 2026-10-05

The IWF's half-year numbers have already passed all of 2025. On October 4–5, 2026, the Internet Watch Foundation reported that between January 1 and June 30, 2026 its analysts assessed 6,310 photorealistic AI-generated images that met the legal definition of child sexual abuse. That is 40% more than the 4,512 recorded in all of 2025. Where age and gender were recorded, girls appeared in 98% of the images. Imagery depicting children aged 7–13 made up 79%, up from 70% last year, and analysts found 190 images of infants and toddlers under two. Category C imagery (sexualised posing and nudity) rose to 88% of AI images, from 62%. The Guardian adds that the UK's Report Remove service has received 420 reports this year from children who believe images of them were faked or manipulated to look explicit, already more than the 397 for all of 2025.

The IWF is using the figures to push the EU to agree its long-delayed Child Sexual Abuse Regulation, including detection of previously unseen material, which hash-matching cannot catch. In the UK, it backs calls for binding AI legislation that would require "safe by design" models.

Context for this file: In March we reported the IWF's earlier finding of a 26,385% rise in AI-generated CSAM videos in 2025. The new figures cover still images only and show the problem growing in volume and shifting toward younger children.

Why it matters for the industry: The pressure is landing on generative-image tools, the same nudification and "spicy" image products at issue in the Minnesota/xAI litigation and in the EU's AI Act amendment. Adult platforms that host AI-generated content should expect regulators to require proactive classifiers, not just hash lists, and age-estimation checks on depicted persons as well as users. The UK's new realistic-AI-image offences, which take effect this autumn (see our Crime and Policing Act file), already point that way.

New Sources