Meta expands AI detection tools for child grooming on Messenger and Instagram

Meta announced on 19 September 2024 that it is deploying expanded artificial intelligence systems to detect adult accounts exhibiting grooming behaviour on Messenger and Instagram, following criticism from child-safety organisations and lawmakers over the volume of abuse occurring on its platforms. The new tools analyse account behaviour including age discrepancies between users in conversation, message frequency and requests to move conversations to other platforms, and will result in accounts flagged as high-risk being restricted from initiating contact with teen users. Meta stated that the AI models were trained on patterns identified in prior investigations and law-enforcement referrals, and that the system is designed to operate without reading message content in end-to-end encrypted chats.

Platform-level controls that restrict who can initiate a conversation reduce the volume of opportunistic contact, but they do not analyse the substance or trajectory of conversations already underway or address harm on the many other services where young people communicate. Guardii, backed by Startmate and operating as a Meta Business Partner, analyses the actual exchange in real time across Instagram, Snapchat, Discord, Roblox and other platforms, using anti-grooming detection to identify coercion, solicitation and the staged escalation tactics that characterise predatory behaviour, and surfaces the threat to the parent, school or safeguarding professional positioned to act before a child is manipulated into offline contact or image sharing.