Age Verification Failures: Age Estimation Locks Out Adults

Picture a parent in Sydney trying to log into an app, getting told she looks too young, and being asked to scan her driver's license to fix it. Now picture that happening to millions of people. That is the quiet problem inside the fight over age verification (proving how old you are before you can use an app), and Australia just carried it into a courtroom.

You are probably worried about two things: "Is this going to make my own life harder?" and "Who ends up holding my ID?" Fair questions. Short answers: possibly, and that depends on choices nobody has made clearly yet. Let's walk through it.

Why Australia's Age Verification Law Is Heading to Court

Australia banned social media for kids under 16. The age requirement is simple to say and hard to do. Communications Minister Anika Wells announced on October 20 that the government is suing Facebook, Instagram, TikTok, Snapchat and YouTube for not complying, according to Reclaim The Net. The maximum penalty is A$109.2 million, roughly $76 million.

That sounds like a government getting tough. But here is the detail that made me sit up. A software test reported by NBC News opened 50 accounts after the law took effect. On none of them did the platform ask for proof of age.

Zero for fifty. So the law exists, the fine is huge, and the front door was still wide open. Sit with that for a second.

What the Australia age verification law actually demands

The age verification rule asks platforms to take "reasonable steps." It does not say which steps. No required accuracy, no named age-verification technology, no pass mark. Meta's response, as quoted by Reclaim The Net, is that age assurance technology (software that guesses or confirms your age) "either doesn't exist or isn't reliable enough to enforce at scale." This article is part of a series, start with Deepfake Voice Scam News Today 77 Of Targets Lost Money.

Age assurance technology "overlooks the practical reality" and "either doesn't exist or isn't reliable enough to enforce at scale."

Meta's argument, as reported by Reclaim The Net

Now, Meta has every reason to say that. It is being sued. I would not take the company's word as gospel. But the other evidence points the same way, which is the uncomfortable part.

---

How Australia Age Verification Can Fail: Kids Slip In, Adults Get Stuck

Every age verification check can go wrong in two directions. A false negative is a kid the system waves through. A false positive is an adult the system wrongly flags as a kid. Most headlines obsess over the first. The people building these systems are finding that the second is the bigger headache at scale, according to a technical analysis from Xident. Keeping kids out was never the hard part. Letting the right adults in is.

Why does that matter to you? Because a wrongly flagged adult has one way out: prove it. That usually means identity verification, a government ID or a more detailed face scan. A broken check does not stay a small annoyance. It turns into a request for your most sensitive details.

Age verification checks compared: what each one gets wrong

Type of check	What goes wrong	Status in the reporting
Document verification for age verification (uploading an ID)	About 3% of results are wrong, either way, in a landmark study covered by ABC News. Stored ID images become a target for hackers and identity fraud.	Enacted as an option, with privacy and security risks
Facial age estimation (software guessing age from your face)	Error margins of 2 to 4 years in that same study. Newer biometric age models reportedly reach about 1.5 years.	Improving, still not precise enough near age 16
No age verification at all	Nothing stops anyone. Zero of 50 test accounts were asked for proof.	Reported failure, now the subject of the court case

Look at that middle row. A 1.5 year miss sounds tiny. But the line here is exactly 16. A miss of a year and a half in either direction means some 17-year-olds look old enough, and some adults look too young. That is not a bug you can patch. It is math. Those numbers come from the ABC News report on the study and the Xident analysis.

Where Does Your Data Go When Age Verification Asks for a Face or an ID?

Here is the straight answer to that question: it depends entirely on the platform, and the rules rarely tell you. Some systems check your face and throw the result away. Others keep ID images or biometric data (your face, voice, fingerprints, the body stuff that is uniquely you). IEEE Spectrum points out that every record kept becomes something a hacker can go after.

That last one is almost funny. Picture a bouncer who lets you try the door again after every rejection. Meanwhile your honest uncle, who just looks young for his age, gets pushed to hand over a passport. Previously in this series: Google Face Recognition 1 Switch Deletes Every Face Code.

Is age verification the same as facial recognition?

No. Age estimation guesses how old a face looks. Facial recognition tries to work out who a person is, and face verification checks whether two photos show the same person. They use related technology but answer different questions. Mixing them up is how people end up trusting the wrong tool for the wrong job.

---

Will Court Fix Age Verification? What the Lawsuit Can and Cannot Do

Probably not the part you care about. A court can decide whether the platforms broke the rule and what the fine should be. It is much less likely to set a technical pass mark for age verification. Reporting says the eSafety regulator already ran into trouble enforcing age verification: existing law did not let it force platforms to hand over their internal documents, which made civil penalty cases hard to win.

So picture the best case for the government. It wins. Platforms pay. And everyone still lacks a clear answer to "how accurate must age verification be, and what must happen to the data afterward?" Companies would then have a strong reason to over-collect, because asking for more ID is the safest way to avoid a fine. Nobody gets fined for asking too much. That incentive is the real risk to ordinary users.

My take: "protect minors" is a goal, not a system. A goal does not tell a platform how to treat a 34-year-old whose face scan says 15.

One thing you can do before trusting an age check

If you have ever wondered whether a photo or a profile is really who it claims to be, that is the very question face technology exists to answer, and it is why accuracy matters so much. So before you scan your face or upload an ID for any app, find one answer: does the company say, in plain words, that it deletes the scan or the ID image once the check is done? If you cannot find that sentence, treat it as a no. And if a check wrongly flags you, ask for a human review instead of simply uploading more documents.

Here is the number I keep coming back to. Zero out of fifty. If age verification that was supposed to be running never ran, how many of us are being asked for our IDs right now by the ones that do?

Age Verification: Frequently Asked Questions