Age Verification Face Scan: 2-Year Miss Lets 16 Pass as 18
Quick answer
How accurate is age verification face scan technology for teenagers?
Not very accurate near the legal age line. Testing by N.I.S.T. found the best systems miss by roughly two to almost three years for people aged thirteen to twenty-four. The software estimates age from facial features rather than knowing it, so a sixteen-year-old can pass as eighteen. Vendors add a buffer age to reduce errors.
Age Verification Face Scan: 2-Year Miss Lets 16 Pass as 18
This episode is based on our article:
Read the full article →Age Verification Face Scan: 2-Year Miss Lets 16 Pass as 18
Full Episode Transcript
A Age Verification Aud 99m Fines Cant Fix A Face Scan Guess scan can look at a sixteen-year-old and decide she's eighteen. According to testing by N.I.S.T., the U.S. government's standards lab, the best age-guessing systems are off by nearly two to almost three years for people between thirteen and twenty-four. That's exactly the range where the legal line sits.
If you've ever taken a selfie to prove your age on
If you've ever taken a selfie to prove your age on an app, this already touches you. And if you're a parent, you may have felt relief when a platform announced it was checking kids' faces. That relief makes sense. It's also a little misplaced.
Platforms like Discord and Roblox now use these scans to decide who sees age-restricted content. The checks take about twenty milliseconds. That speed feels like precision. So how does a system that fast still miss by two years?
Start with what the software is actually doing. It isn't recognizing you. It doesn't know who you are. It looks at skin texture, jawline, and eye shape, then makes a statistical guess at your age. That's an estimate, not a fact.
Why does the guess drift? Because faces between thirteen and twenty-four are still changing, and those features overlap a lot from one year to the next. A ten-year-old and a forty-year-old are easy to tell apart. A sixteen-year-old and an eighteen-year-old can look nearly identical to a machine.
People assume a fast check is an accurate check
People assume a fast check is an accurate check. That's understandable, because platforms advertise the speed loudly, and we're trained to link "instant" with "smart." But speed here means the model does a quick pattern match. It trades precision for convenience.
Independent testing of Roblox's face scan found estimates off by about one and a half years on average for users under eighteen. According to that same reporting, the system has processed around three hundred thirty-eight million age checks. Even a small share of misses at that scale means a huge number of kids slipping through.
For platforms, that's a compliance problem. For a parent, it means the gate on your child's phone may be thinner than it looks.
So what do companies do about the error? They build in a cushion. To confirm someone is over eighteen, a service may set the bar at twenty-five. If the system thinks you look twenty-five or older, you pass. Otherwise, you get sent to another check.
Yoti, one of the major vendors, reports that with
Yoti, one of the major vendors, reports that with that twenty-five cushion, the error rate for thirteen to seventeen-year-olds is about one in a thousand. Lower the cushion to twenty-two, and it climbs to about seven in a thousand. So the accuracy comes from the cushion, not from the machine being sharp.
There's a cost. Real adults between eighteen and twenty-four get bounced and have to hand over an ID instead. The system stays safer by turning away people it shouldn't.
Research also points to bias. One study of Microsoft's Azure found better accuracy for male faces than female ones. Image quality, expression, and facial hair all shifted the results. Two teenagers, same lighting, same camera, could get different outcomes.
And then there's the simplest problem. Almost half of minors online admit they've lied about their age to make an account. A face scan doesn't end that. It just moves the game to a new set of tricks.
The Bottom Line
The technology works best where it matters least, and worst exactly where it matters most. It can easily tell a child from an adult. But the sixteen-year-old standing right at the line is the face it's most likely to get wrong.
A face scan doesn't know your age. It only guesses. The guess is good for obvious cases and shaky near the legal line, which is where teenagers live. So a scan can be one layer of protection, but it can't be the whole wall.
If that unsettles you, good. It should make you ask more of the platforms your kids use. Understanding the gap is how you stop feeling powerless about it.
Check the show notes for the full piece.
Ready for forensic-grade facial comparison?
Full forensic reports with detailed similarity scoring. Results in seconds.
Run My First SearchMore Episodes
Biometric Camera: A Facial Match Is a Lead, Not Proof
Police in the United States have wrongfully arrested at least fourteen people because a facial recognition match turned out to be wrong. Now Scotland is considering this same technology, and it may do so without a single
PodcastInstagram Age Verification: AI Can Lock Adults in Teen Mode
Meta says it's now using A.I. to guess the age of Instagram users around the world. If the system decides you're a teenager, it can lock your account into teen mode. That's true even if you typed in an
PodcastMinimum ACT Score: Colleges Vary, EU Social Media Age Is 15
Europe's own app for checking a child's age online was cracked in under two minutes. It was the Commission's showpiece, built to prove that you could protect kids and protect privacy at the same time.
