Thorn Releases Research on Generative AI and Child Sexual Exploitation

Child safety organisation Thorn has published research examining the use of generative AI in child sexual exploitation, finding that offenders are using AI image generators to create synthetic CSAM, produce fake explicit images of real children from innocuous photos, and generate text to refine grooming scripts. The study, based on analysis of dark web forums and interviews with law enforcement, found that 68 percent of surveyed investigators had encountered AI-generated CSAM in the past year, and that offenders are sharing prompts designed to bypass content filters. Thorn warns that generative AI lowers the technical barrier to producing exploitative material and accelerates the grooming process by enabling offenders to generate persuasive, personalised messages at scale.

AI that enables offenders demands AI that stops them, and Guardii's anti-grooming, anti-sextortion and anti-CSAM detection modules are engineered to recognise both the material Thorn describes and the conversations in which it is shared or solicited. The system operates across Instagram, Snapchat, Discord, Roblox and other platforms, identifying the behavioural signature of exploitation whether the offender is using a script, a generative prompt or their own words, and whether the image is photographic, synthetic or deepfake. By flagging sustained targeting, escalation and coercion in real time, Guardii closes the operational gap Thorn identifies: it intercepts the grooming conversation before the AI-generated content is deployed, surfaces the child at risk to a parent or safeguarding professional, and enables law enforcement to intervene while the offender is still in contact.