Osegi
Security·2 min read

Childlight 2026 index: AI child abuse reports up 8,411% in two years

Reports of AI-generated child sexual abuse material jumped from about 4,700 in 2023 to over 400,000 in 2025. The number counts reports, not victims.

A parent and child looking at a tablet screen together in a dim room, illustrating online child safety concerns.
Image: البوابة التقنية

Reports of AI-generated child sexual abuse material rose from roughly 4,700 in 2023 to more than 400,000 in 2025, according to the 2026 "Into the Light" index from Childlight, a child safety institute at the University of Edinburgh. That is an increase of 8,411% in two years.

The figure comes from reports that tech companies file with the US National Center for Missing and Exploited Children (NCMEC). It counts reports logged, not children harmed and not proven crimes.

Does the 8,411% rise mean abuse grew by the same amount?

No. Childlight stresses that the number is not a count of victims. Detection systems and reporting routines change over time, so better tools for spotting this material can explain part of the jump, alongside a real rise in how widespread it is. The Guardian's coverage on 16 November made the same point about improved detection.

What the volume does show is how much work now lands on the people who review, remove and respond to this content.

What counts as AI-generated abuse material?

The index covers images and videos created or altered with AI tools. That includes deepfakes, which can turn real photos of a child into abusive content. Such material can be used to harm, threaten or blackmail children, even when the images are fabricated.

Where the data comes from

This is the third edition of the index. It draws on NCMEC, the UK's Internet Watch Foundation and the INHOPE network, and covers reported material, reported abuse, survey results and protection services. Childlight notes that these sources differ in scope, definitions and monitoring methods, so report counts should not be read as a full estimate of victims.

New regional survey data covers North America, East Asia and the Pacific, and Latin America and the Caribbean, countries that hold about 63% of the world's children. Even so, US report data is not a direct measure of how common these crimes are in any other country.

The institute calls for safeguards that stop image generators being misused, stronger detection and removal on platforms, closing legal gaps around synthetic material, and support for specialist reporting hotlines. The 2026 data is available in the University of Edinburgh's repository, published on 16 November.

Apps worth a look

SUMO: the number puzzle

Tap the right tiles, add them up, hit the target. Easy to learn, hard to master. Free, for all ages.

Tags:childlightchild safetyai-generated contentdeepfakesncmecuniversity of edinburgh

More from this day

See all →

More in: Security

See all →