New research reveals AI-generated CSAM proliferating on mainstream platforms

A study published in early September 2026 by the Stanford Internet Observatory and the National Center for Missing and Exploited Children has documented a rapid increase in AI-generated child sexual abuse material circulating on Instagram, Discord, Telegram, and other widely used platforms. Researchers identified more than 3,400 accounts sharing synthetic images created using generative-AI tools, many of which depicted recognisable features scraped from photographs of real children posted by parents or harvested from school and sports websites. The report notes that current content-moderation systems, which rely heavily on hash-matching against known abuse imagery, are largely ineffective against novel AI-generated material, and that offenders are using these images both for distribution and as part of sextortion and grooming campaigns to desensitise or coerce minors.

Guardii's CSAM detection capability, which includes a dedicated module for AI-generated and deepfake abuse material, addresses precisely the gap this research exposes. The world-leading platform uses behavioral and visual pattern recognition rather than static hash databases, enabling it to identify previously unseen synthetic imagery in direct-message exchanges across Instagram, Snapchat, Discord, Roblox, and other services. When AI-generated CSAM is detected in a conversation, Guardii blocks the sender, alerts the intended recipient's parent or safeguarding authority, and enables immediate escalation to law enforcement. Because the system evaluates conversational context and escalation alongside image content, it can distinguish sextortion or grooming sequences that incorporate synthetic material from other activity, intercepting both the distribution of novel abuse imagery and the coercion it facilitates before a child is harmed.