Study reveals alarming rise in AI-powered child abuse deepfakes

The latest report from the Internet Watch Foundation (IWF) highlights a worrying rise in the creation of “AI-generated child sexual abuse images.” Advanced technology now allows pedophiles to add “another person’s face or image to a real video” in a near-seamless manner, resulting in “highly realistic deepfakes.”

Criminals on the dark web are boasting about the ability to “create any child pornography we desire (…) in high definition,” according to the IWF report, which identified the “first compelling examples” of AI videos depicting child sexual abuse. Published on Monday (22), the report by the organization, which monitors the global circulation of Child Sexual Abuse Material (CSAM), highlights that the production of AI-made videos is increasing as AI tools become more accessible and easy to use.

Increasing quality and realism

Deepfake videos, or videos partially manipulated by AI, are becoming increasingly realistic, featuring faces — sometimes of real victims, sometimes of different people — in scenes that appear authentic. In 2023, when the IWF first spotted AI-generated content, the quality was basic. Now, Dan Sexton, the IWF’s chief technology officer, says future videos will have “higher quality and realism.”

The IWF has identified the first fully synthetic videos of child sexual abuse, indicating that “offenders replicated and shared AI models with over 100 abuse victims.” On a dark web forum dedicated to child sexual abuse material, investigators found nine deepfake videos circulating between March and April of this year.

The report explains that “some of the deepfake videos feature adult pornography that is altered to show a child’s face. Others are existing videos of child sexual abuse that feature another child’s face superimposed on them.” Because the original videos involve real children, the deepfakes are particularly convincing.

An 18-second AI-generated video found on the clear web shows an adult man abusing a girl who appears to be around ten years old. Despite flaws in the video, IWF analysts describe the activity as clear and ongoing.

A dark future

Susie Hargreaves OBE, CEO of the IWF, highlights that “generative AI technology is evolving rapidly, allowing offenders to produce graphic videos of child sexual abuse at will, which is quite terrifying.” She emphasises that “the fact that some of these videos manipulate images of known victims is even more terrifying. Survivors of some of the worst types of trauma now have no respite, knowing that offenders can use footage of their suffering to create any abuse scenario they desire.”

Hargreaves warns that “without proper controls, generative AI tools provide a playground for online predators to play out their most perverse and disgusting fantasies. Even now, IWF is beginning to see more of this type of material being shared and sold on commercial child sexual abuse sites across the internet.”

Via News Agencies