Social media platforms have spent years telling parents their children are safe online. New research suggests those assurances don’t hold up. A report from the Cybersafety Research Center tested 86 child safety features across TikTok, Instagram, Snapchat, and YouTube. Only 35 worked as promised, and the rest were broken, buried in settings, or missing entirely.
Which social media platforms performed the worst on child safety?
To run the tests, researchers created fake teen accounts and adult accounts to see whether safety features worked in practice. Snapchat had the worst failure rate at 73%, followed by Instagram at 66%, YouTube at 55%, and TikTok at 50%. Every conduct safeguard designed to prevent cyberbullying failed across all four platforms.
On TikTok, a minor’s test account searching for content related to disordered eating was met with the app’s own suggestions for terms linked to pro-anorexia communities, including phrases about hiding food and self-harm.
On Snapchat, an adult test account was able to find and message a child account without any restrictions at all. Meanwhile, Instagram prevented adults from starting conversations with teens who didn’t follow them, but once a child messaged an adult first, that adult could reply freely with no warnings.
Across all four platforms, nine features were classified as completely missing, meaning researchers could not trigger them even after following the steps each company described.
How the platforms responded and what this means for child safety online
All four companies disputed the findings, arguing their features work as intended or that the tests didn’t reflect how real kids use the apps. These findings come as the UK moves toward a social media ban for under-16s, while similar restrictions gain traction in other countries.
Separate research has also found that Australia’s outright ban on under-16s has not stopped 85% of teens from accessing social media anyway, as kids have proven surprisingly creative at bypassing age checks altogether.
The bigger problem is getting harder to ignore. If platform safeguards are weak and bans are easy to dodge, your child may be relying on systems that are far less safe than they seem.