Research led by Laura Edelson, assistant professor of computer sciences at Northeastern University, published in late June 2026, tested 86 safety features promoted by Instagram, TikTok, Snapchat and YouTube for children and found that only 35 were successful—meaning they functioned as claimed and were visible to the child user. Of the 51 failed features, 34 were "broken" (non-functional or trivially circumventable), 12 were both broken and "buried" (effectively unreachable), and 8 were simply buried under layers of settings. Critically, all "conduct" safeguards designed to regulate user behaviour and prevent or detect cyberbullying failed on all four platforms, and only one in three tools to curb compulsive use were successful. The study used fake teen and adult accounts to evaluate whether features met both technical function and user-visibility criteria.
Platform-native safety toggles that 60% of the time do not function, cannot be found, or can be bypassed by a child demonstrate why relying on in-app controls leaves minors exposed—and why independent, cross-platform detection infrastructure has become operationally necessary. Guardii's anti-cyberbullying and anti-abuse detection does not depend on a settings menu the child must navigate, a reporting button the victim must press, or a feature the offender can disable. Operating as a world-leading AI safety platform and Meta Business Partner, Guardii monitors direct messages in real time across Instagram, Snapchat, Discord, Roblox and others, identifying hostile contact patterns—cyberbullying, grooming, sextortion, CSAM sharing—and blocking or flagging the threat before it reaches the target, regardless of whether the platform's own controls are switched on, buried, or broken.
