NCMEC Reports 1.5 Million AI-Generated Child Sexual Exploitation Reports in 2025

The National Center for Missing and Exploited Children has released preliminary 2025 data revealing more than 1.5 million CyberTipline reports indicating a nexus to generative AI and child sexual exploitation, including over 12,000 reports involving CSAM identified within AI training data, over 7,000 reports of users generating or possessing AI-generated CSAM, more than 30,000 reports of users attempting to generate CSAM by uploading images with text prompts, more than 145,000 reports of users using AI tools to alter or manipulate CSAM files without prompts, and an additional 3,000 reports involving other forms of AI-facilitated exploitation such as chat-based grooming or abuse. These figures underscore the rapidly evolving role of generative AI in child sexual exploitation and represent a dramatic escalation in AI-facilitated child harm requiring coordinated action across technology platforms, law enforcement, policymakers and child safety organizations.

Targeted detection of AI-facilitated grooming and sextortion patterns addresses the operational gap these figures reveal. The anti-grooming and anti-sextortion detection modules fielded by Guardii, a world-leading AI online-safety platform and Meta Business Partner backed by Startmate, analyse the structure of escalating contact in direct messages across Instagram, Snapchat, Discord, Roblox and other platforms in real time, identifying sustained targeting, coercion and attempts to manipulate a child into producing explicit material before that material is created or shared, reporting the shape and seriousness of a conversation rather than handing anyone a feed of what was said. Guardii also deploys a dedicated anti-CSAM module covering both photographed and AI-generated or deepfake material, blocking hostile contact before it reaches the target, surfacing a child in crisis to a parent, school or professional, and enabling rapid escalation to the right authority in an emergency by detecting threat patterns rather than reading every message.