Meta Tries to Stop Kids from Using Fake Mustaches to Bypass Age Checks

Meta announced new tools on May 5 to better identify when children use its platforms underage. The timing is notable because a recent study shows just how easy it is for young people to fool these checks—sometimes with nothing more than a pencil drawing.

How Meta's New System Works

Meta has built what it calls an "adult classifier," a tool that uses artificial intelligence to guess whether someone is older or younger than 18. Instead of relying on a single check, it looks at multiple clues: what people post, what they comment on, what they write in their bios, and even what their photos look like.

The company says its system analyzes photos to spot physical characteristics like height and body shape that might suggest age. Importantly, Meta says it does not use facial recognition—the technology that identifies you by your face specifically. It uses broader visual cues instead.

Previously, Instagram had tested a different system that looked at people's faces in video selfies to guess their age. The new approach casts a wider net.

The Problem: Kids Find Workarounds

A study by Internet Matters, a UK research organization, shows how well these systems actually work. The results are not encouraging. About 32% of UK children said they had successfully fooled an age check. Nearly half—46%—believed these checks are easy to get around. Even more surprising: 16% of parents actively help their kids bypass these systems.

The methods vary. Some children use VPNs, which hide where they are online. Some log into their parent's account. Others use simpler tricks.

The most striking finding involved a 12-year-old who drew a fake mustache on their face with eyebrow pencil. When they submitted a photo with the drawn-on mustache, the system guessed they were 15. The system had been fooled by a pen drawing.

This reveals something important about how visual age-checking actually works. These systems spot visual patterns associated with maturity—like facial hair—rather than using something more fundamental to identify someone. A simple marker line was enough to shift the system's guess upward by years.

The Broader Picture

The UK government passed the Online Safety Act in 2023, which requires social media platforms to protect children from harmful content. That law puts age verification front and center. Platforms now face real legal pressure to keep underage users away from adult content.

However, there is a tension here. The better you want to know someone's real age, the more personal information you need to collect. That raises privacy concerns for everyone. Meta's choice to avoid facial recognition is partly about this tradeoff—they want to check age without collecting detailed biometric data that could be misused.

We have seen similar patterns before in internet history. When spam filters first arrived, people found ways around them. When websites added paywalls, users developed workarounds. The same thing is happening now with age verification. New tools prompt users to find new ways to bypass them. It becomes an arms race.

Why This Matters

The research also found something worth thinking about: when 16% of parents actively help their children get past age checks, no technology alone can solve the problem. Part of the issue is social—it is about what families accept and allow—not just what software can catch.

Meta's new multi-method approach makes sense on paper. By combining visual clues, behavior patterns, and what people post, you get more information than any single check provides. It is similar to how banks use multiple forms of ID to verify who you are. But the fake mustache story shows the risk: if the system relies too heavily on surface-level visual markers, it can be fooled by simple tricks.

What Comes Next

As governments push platforms harder to verify age, and platforms invest more in the technology, this cat-and-mouse game will likely continue. Kids and parents will find new workarounds. Platforms will add new detection layers. Meta's approach is a step forward, but the research suggests there is still a long way to go before these systems work reliably while still protecting people's privacy.