Bloomberg investigation reveals open-source AI tools fuel rise in AI-generated child sexual abuse material with limited tracking capability

A Bloomberg investigation published 6 October 2026 documents how open-source AI models are fueling a rise in explicit AI-generated child sexual abuse material because, unlike commercial cloud-based AI tools such as ChatGPT and Claude which operate under safeguards, open AI models can be downloaded and modified to produce illegal content with almost no ability for authorities to track usage once the models leave the server. A forum dedicated to AI-generated CSAM stated it no longer allows topics asking how to use online AI generators and now supports only local offline AI generation, demonstrating a deliberate shift to evade detection. The article notes that current laws are ill-equipped to address the accessibility of these downloadable tools, which enable technically capable offenders to create unlimited illegal images.

Commercial platforms may impose upstream guardrails but the proliferation of synthetic CSAM means detection must operate at the point of contact and distribution, where Guardii's anti-CSAM detection module, which includes specific capability for AI-generated and deepfake material, can identify and block images in direct-message exchanges before they reach a child or are shared within a network. Backed by Startmate and operating as a Meta Business Partner, Guardii detects the distribution pattern itself rather than relying on a centralised model to refuse a generation request, addressing the enforcement gap created when open-source tools place illegal image creation beyond the reach of any single platform or regulator and surfacing distribution networks in real time for escalation to law enforcement.