Age Verification: Adults Flagged as Kids Get Pushed for ID

Age Verification: Adults Flagged as Kids Get Pushed for ID

Full Episode Transcript

---

Australia's government says it opened fifty new social media accounts after its under-sixteen ban took effect.

---

A software test found that not one of those

A software test found that not one of those accounts was asked to prove an age.

So here's a question for anyone who's ever scrolled past a pop-up asking how old you are.

What happens when the system built to catch kids starts catching grown adults instead?

If you've ever opened a social media account, this story is about you.

Soon, a camera or a request for your driver's license may stand between you and your own account.

Here's the situation.

Communications Minister Anika Wells announced on October twentieth that Australia is suing Facebook, Instagram, TikTok, Snapchat, and YouTube.

The government says they failed to comply with the country's ban on social media for children under sixteen.

The maximum penalty is now about one hundred nine million Australian dollars.

Meta argues that the technology to check ages either doesn't exist or isn't reliable enough at scale.

So the real question is this.

If age checks can't be both accurate and gentle, who pays for the mistakes?

---

Start with that software test

Start with that software test.

It's reported by NBC News.

Testers created fifty accounts after the law took effect.

The platforms never asked for proof of age.

That's not a case of weak checks.

It's a case of no checks at all.

Now the government wants to prove that in court.

That turns out to be hard.

According to the article, the regulator, called eSafety, found that existing law didn't let it force platforms to hand over their internal documents.

The regulator had already warned that civil penalty cases were unlikely to succeed.

And even a win in court wouldn't set a technical standard.

The law says platforms must take "reasonable steps."

---

It never says what those steps are

It never says what those steps are.

That's a bit like telling drivers to go a reasonable speed, with no signs on the road.

That leaves platforms with three bad options.

Do nothing and risk a lawsuit.

Guess ages from your face and risk misjudging adults.

Or demand ID documents and risk a privacy disaster.

For companies, that's a legal puzzle.

For you, it means the choice may be a selfie scan or handing over your passport just to read your own messages.

Let's look closer at the Korean Deepfake 17 378 Sexual Violence Cases In 8 Months guessing.

Facial age estimation means software looks at a face and predicts how old the person is.

The article says leading models have cut their average error to about a year and a half by early twenty twenty-six.

Now picture it at the legal line of sixteen.

---

A year and a half of wiggle room means someone

A year and a half of wiggle room means someone who's seventeen could look fifteen to the software.

It also means a fourteen-year-old might pass as sixteen.

An earlier Australian study, covered by the A.B.C., found bigger gaps.

It put error margins for facial estimation at two to four years.

Document checks did better, with roughly three percent mistakes in either direction.

Three percent sounds tiny.

But across millions of users, that's a lot of real people locked out.

That's the part the article says teams are now discovering.

The hardest job was never keeping kids out.

It's letting adults in.

Engineers call an adult wrongly flagged as a child a false positive.

For that adult, it means being told to upload an ID after the machine decided they look too young.

---

That ID has to go somewhere

And that ID has to go somewhere.

Platforms that store face scans and ID images create new targets for hackers.

Every saved record is one more thing that can leak.

So the fix for one problem can create another, and you carry the risk.

Kids, meanwhile, find ways around it.

According to IEEE Spectrum, some platforms let users retry the same check until it passes.

Others get fooled by borrowed IDs, V.P.N.s, or cycling through accounts.

So the adult gets stopped, and the determined teenager gets through.

Some people say bigger fines will force better technology.

But a fine can't shrink a margin of error.

The real question isn't how hard regulators punish platforms.

It's which mistakes they decide to notice.

---