AI child abuse imagery surges 40% as IWF demands EU regulation

The rapid proliferation of synthetic illegal content exposes a critical gap in current moderation technologies that rely primarily on hash-matching for known material. Because AI tools allow for the creation of unique, previously unseen imagery, platforms require new legal frameworks to deploy proactive detection tools without facing liability for privacy violations. For the broader streaming and hosting ecosystem, this shift signals an impending transition from reactive content takedowns to mandatory, algorithmic scanning requirements. Industry stakeholders should monitor the final text of the EU Child Sexual Abuse Regulation to see if it mandates the detection of 'unknown' content, which would necessitate significant investment in advanced computer vision stacks.

The Internet Watch Foundation has been building a comprehensive evidence base to pressure EU legislators. In March 2026, the IWF published its report "Harm without limits" documenting the full scale of AI-generated child sexual abuse imagery discovered by its analysts, revealing a 260-fold increase in AI-generated videos and calling for a comprehensive EU ban on all AI child sexual abuse content and the tools used to create it. The report recommended mandatory pre-market assessment of AI systems, safety-by-design requirements under the EU AI Act, and use of the IWF's Hash List of over 3 million verified CSAM hashes to block known imagery from training data. On the regulatory front, the IWF has been tracking two parallel EU legislative tracks. The organization welcomed the political agreement reached on 23 June 2026 on the recast Child Sexual Abuse Directive, which criminalises the design, adaptation, distribution, and supply of AI systems built specifically to generate child sexual abuse material as a minimum standard across the EU. The Directive also introduces criminalisation of possession of AI-generated child sexual abuse material and instruction manuals that facilitate abuse. However, the IWF has separately warned that the EU's failure to extend the temporary derogation to the ePrivacy Directive creates legal uncertainty for platforms voluntarily detecting child sexual abuse material, meaning technology companies face questions about whether they can search for and block such content on their services without explicit legal cover. The IWF's operational model relies on converting confirmed criminal images into unique digital fingerprints called hashes, which are shared with law enforcement and technology companies for automated detection. The organization's 2026 report details how its analysts have been assessing imagery depicting children under two years old, underscoring the severity of content being generated. The IWF supports the European Parliament's position on the Directive, including provisions targeting CSA image-generators, and has called on Member States to transpose the legislation swiftly once formally adopted.