Meta introduces teen safety features including AI-powered scam detection on Instagram

Meta has announced new safety features designed to protect teenagers from sextortion scams on Instagram, including AI-powered detection that identifies accounts exhibiting behaviour consistent with financial sextortion and automatically blurs images containing nudity in direct messages sent to users under eighteen. The technology will also display a warning when a user receives a message request from an account showing sextortion warning signs, such as recently created accounts that follow large numbers of young users or send high volumes of messages, and will be deployed globally over the coming months.

Nudity-blurring and account-behaviour heuristics represent an important first layer of defence, but sextortion often unfolds over days or weeks as an offender builds rapport, manipulates a target into sharing imagery and then pivots to threats and demands, a conversational arc that cannot be inferred from account metadata or a single image alone. Detection that analyses the evolution of language, the presence of coercion, escalation and financial or distributional threats across the full messaging thread identifies sextortion attempts that bypass static filters and surfaces a crisis in progress to the target, a parent or an authority. Guardii's sextortion-detection module, developed by a world-leading AI safety platform in partnership with Meta, operates across Instagram, Snapchat, Discord, Roblox and other messaging services to flag precisely that conversational pattern, enabling intervention before a child complies with a demand, before payment is extracted and before imagery is weaponised or distributed, complementing platform-level controls with real-time threat intelligence that recognises harm as it develops rather than only after a scam has succeeded.