AI fuels surge in child abuse imagery, watchdog warns

Fresh Euronews report examines a sharp rise in artificial intelligence-generated child sexual abuse material, highlighting growing concerns among investigators and policymakers about the scale, sophistication and accessibility of such content.

Citing new findings from the Internet Watch Foundation (IWF), the article states that AI-generated imagery of child abuse increased by 14% in 2025. The organisation identified more than 8,000 such images and videos through user reports over the past year. Although this remains a relatively small proportion of the total material the IWF monitors, the report stresses that the growth rate is significant and accelerating.

A central concern outlined in the original reporting is the increasing realism of AI-generated content, which is making it harder to distinguish from genuine photographs. The IWF classifies material as AI-generated based on visible inconsistencies, admissions by victims, or disclosures from creators about the use of AI tools. However, as these technologies improve, such indicators may become less reliable.

The data suggests that AI is not only expanding the volume of abusive material but also its severity. Of the AI-generated content identified, more than 3,400 items were “full-motion” videos described as highly realistic and capable of depicting multiple participants. The report found that over 65% of these videos fell into the most serious legal category of abuse under UK law. By comparison, 43% of non-AI-generated material was classified at this highest level, indicating that AI is being used to produce more extreme content.

The report also sheds light on how offenders are adapting to technological change. According to the IWF, individuals are using online forums, including parts of the dark web, to share knowledge, tools and datasets. In some cases, users are offered training on how to generate synthetic images, while increasingly accessible software allows content to be created with minimal technical expertise. The report notes that some systems require only a single reference image to generate further material.

While many users are producing relatively simple content, the IWF identified a smaller group of more advanced creators responsible for longer and more complex productions. In one cited example, a creator received thousands of endorsements for producing a lengthy AI-generated video. However, the organisation cautions that its findings represent only a partial picture, as investigators are unable to access material behind paywalls or within encrypted platforms.

The report has prompted renewed calls for regulatory action. The IWF’s chief executive urged the European Union to introduce a bloc-wide ban not only on AI-generated abusive material but also on the tools used to create it. Such measures, the report suggests, should apply even to privately generated content.

The issue is being considered alongside broader legislative efforts. European lawmakers have recently extended provisions under the ePrivacy Directive, allowing platforms to continue detecting abusive material while a longer-term framework is negotiated. At the same time, there are calls to strengthen the EU AI Act by classifying systems capable of generating such content as “high risk,” subjecting them to stricter controls before deployment.

Overall, as reflected in the Euronews report, the rapid evolution of AI tools is reshaping the landscape of online abuse, presenting new challenges for enforcement and raising urgent questions about regulation, prevention and technological accountability.

By Tamilla Hasanova