AI Child Abuse Images Surge As IWF Urges EU Action

The Internet Watch Foundation says it identified more criminal AI-generated child sexual abuse images in the first six months of 2026 than in the whole of last year. Its analysts assessed 6,310 such images between January and June, compared with 4,512 images across 2025. The UK-based child protection group is using those figures to press European lawmakers for stronger, lasting rules on detection and removal.

Those numbers, released by the IWF on October 5, describe material its analysts found and judged to meet the legal definition of child sexual abuse. They are not an estimate of every image circulating online. Even with that limitation, the comparison is stark: six months have already exceeded the previous full year’s count by about 40 percent.

The harm is not erased because software generated an image. Some material is created by manipulating photographs of real children, which can retraumatise victims and give offenders another way to target them. Other images are synthetic but still contribute to a market for abuse. The IWF says the speed at which new material can be made is testing systems designed for a slower-moving problem.

One common defense is hash matching. Once analysts confirm a criminal image, they can create a digital fingerprint that helps platforms detect known copies without repeatedly viewing the original. That is valuable, but it cannot identify a newly generated image before someone has reviewed and fingerprinted it. The IWF argues that a framework limited to material already known to authorities would leave a growing gap.

The organisation wants the EU’s proposed Child Sexual Abuse Regulation to give services durable legal certainty to find and remove both known and previously unseen criminal content. That goal has to be pursued with clear safeguards, accountability and proportionate methods. Child protection and privacy are not questions to be waved away with a slogan; the practical design of detection systems matters.

The new data comes amid broader concern about how easily image-generation tools can be misused. Recent restrictions on AI image features show that platforms are being pushed to address risks to minors at the product level as well as through moderation after content appears. Prevention, reporting, victim support and lawful detection all have roles; none is a complete answer alone.

The IWF’s count also captures a change in scale that policymakers cannot ignore. Offenders no longer need to find and redistribute only existing images; generative tools can produce new ones rapidly, frustrating systems built around matching known files. The group says this is why Europe needs a response that keeps pace with current technology rather than the conditions that existed when the legislation was first proposed in 2022.

A measured response begins by taking the verified figures seriously without pretending they describe the entire internet. The rise from 4,512 images in a year to 6,310 in half a year is enough to show why delays have consequences. What lawmakers now have to decide is how to equip services to interrupt that growth while protecting the rights and safety of everyone using them.