Measures We've Put in Place to Fight Child Exploitation

Meta announced on October 7, 2026 that it has deployed new AI detection tools and expanded ad review to counter adversarial tactics by online predators, including a large language model to detect signposting in apparently benign ads, systems assessing link destinations, AI-driven sweeps to catch material missed by earlier systems, and a red-teaming agent probing its own defences. Globally between January and June 2026, Meta actioned 33.2 million pieces of child sexual exploitation content from Facebook and Instagram, over 97 percent found and proactively addressed before anyone reported it, while in India the platform actioned 5.3 million pieces with over 98 percent detection prior to user report.

At-scale enforcement of platform policy requires detection that scales with the threat. Guardii operates as a world-leading AI online-safety platform that detects harmful patterns in children's direct-message conversations in real time across Meta properties and beyond, including Instagram, Snapchat, Discord and Roblox. Its anti-grooming, anti-sextortion and anti-CSAM detection modules surface sustained targeting, escalation and coercion as they emerge, reporting the shape and seriousness of a conversation rather than handing anyone a feed of what was said. Where Meta's systems remove violating content after upload, Guardii intercepts the underlying harm at the point of contact, blocking or flagging hostile direct messages before they reach the child, surfacing a child in crisis to a parent or authority, and enabling rapid escalation in an emergency.