AI-Generated Child Abuse Images: A Growing Threat

Two years ago, when criminals began using artificial intelligence (AI) to generate images of child sexual abuse victims, the representations found by the Internet Watch Foundation (IWF) were still crude, although their cruelty was no less disturbing than that of traditional images.

Since then, the threat has worsened considerably. Thanks to rapid technological advances, the most convincing AI-generated content can now be visually indistinguishable from real images and videos. This is all the more alarming because they can be produced on an astonishing scale and at an astonishing speed. A single instruction can generate at least 50 images, each taking barely twenty seconds to create. Furthermore, this technology is widely accessible, allowing almost anyone to surreptitiously create high-quality and deeply unhealthy content.

In some cases, existing images of children being sexually abused have been used to train AI models, integrating real trauma into artificial content. Predators thus manipulate images showing violence that has occurred to adapt them to their fantasies and preferences. This prolongs the victim’s suffering, as these new stagings can be shared endlessly.

New Avenues for Predators

Research shows that there is a real and undeniable link between viewing child sexual abuse and acting on it. A study of dark web users revealed that 40% of offenders said they would seek to contact a real child after viewing some form of sexually abusive image.

By delving into a dark web forum, IWF analysts discovered that more than half of the AI-generated child pornography images depicted children of primary school age (7 to 10 years old) and that 143 images showed children aged 3 to 6 years old, and two of babies. More than 20% of these images were classified as Category A under UK law (the most serious, involving rape, torture or bestiality).

---

Enjoyed this post by Thibault Helle? Subscribe for more insights and updates straight from the source.