A report published by the Stanford Internet Observatory on 13 August 2024 found that AI-generated child sexual abuse material is increasingly appearing on mainstream social and messaging platforms, evading traditional hash-based detection systems because the images have no known precedent in law enforcement databases. Researchers documented instances on Discord, Telegram and lesser-known forums where users share prompts and tools to generate photorealistic synthetic imagery, with some content subsequently weaponized in sextortion schemes against real children.
Hash-matching depends on prior identification; synthetic material generated on demand presents no existing fingerprint. Anti-CSAM detection that inspects visual and contextual markers in real time—including AI-generated and deepfake variants—can identify novel abuse imagery as it is shared in private channels, blocking transmission before the file reaches its intended recipient. Guardii's CSAM filter applies this logic across platforms, surfacing not only known material but also algorithmically generated content that exhibits the structural hallmarks of abuse, and enabling immediate escalation to the National Center for Missing and Exploited Children or equivalent authorities.
