Thorn releases 2024 research on AI-generated child sexual abuse material
Guardii Analysis
Child safety organisation Thorn has published new research documenting the accelerating production and distribution of AI-generated child sexual abuse material, reporting that offenders are using generative AI tools to create photorealistic imagery of children and that the volume of such material detected by hash-sharing consortia has increased significantly in 2024. The study finds that synthetic CSAM is increasingly difficult to distinguish from photographed abuse, is being shared on mainstream platforms and dark-web forums, and in some cases is being used as part of grooming strategies to manipulate children into believing that producing explicit imagery is normalised behavior.
Detection of AI-generated CSAM cannot rely on hash-matching of known imagery because each synthetic output is unique. Guardii's anti-CSAM module, part of a world-leading AI safety platform backed by Startmate, applies behavioral and content analysis to flag sharing of child sexual abuse material in real time, whether photographed or synthetically generated, and integrates with anti-grooming detection to identify offenders using fabricated imagery as a manipulation tactic in direct messages. By analysing both the material and the conversational context in which it appears across platforms including Instagram, Snapchat, Discord and Roblox, the system surfaces threats that evade traditional hash databases and enables rapid escalation to law enforcement.