Social Media Age Verification Laws: 3 Checks, 3 Data Risks

Seventy thousand people uploaded a photo of their government I.D. to Discord, just to prove how old they were. In a breach last year, those images got out. Now Discord is rolling out age checks again. This time, for most people, it says it won't ask for an I.D. at all.

If you or your kids use social media, the apps may

If you or your kids use social media, the apps may soon need to know how old you are. That means a selfie, a document, or a quiet analysis of how you behave online. It's a lot to hand over, and it's reasonable to feel uneasy about it. According to reporting from TechCrunch and Yahoo Tech, Discord's new age verification began on 09/23/2026. For about nine out of ten users, Discord says it'll estimate age from behavioral signals instead of scanning an I.D. Meanwhile, France is moving to keep kids under fifteen off social media entirely. And here in the U.S., about half the states either have age laws on the books or are working on them. Protecting kids online is the easy part to agree on. So the real question is this, what does it cost everyone else to prove they're not a kid?

Start with that breach. Every I.D. an app collects is a record that has to be stored somewhere. Your name, your birthday, your face, sometimes your home address. Discord's leak showed that this risk isn't theoretical. And that leak is a big reason platforms are now scrambling for methods that skip the document entirely.

One option is the selfie. You look into your camera, and software guesses your age from your face. No I.D. required. But how good is the guess? The National Institute of Standards and Technology, known as N.I.S.T., tested these systems on people between thirteen and twenty-four. On average, the software missed by roughly two years. Some systems missed by closer to three. Three years is the whole gap between a fifteen-year-old and an eighteen-year-old.

That error cuts both ways. A young teen could slip through as an adult. And a nineteen-year-old college student could get locked out of an app she's legally allowed to use. Regulators are watching that fairness problem closely. For everyone else, it means your Facial Recognition Glasses 1 Glance Pulls A Home Address might tell a machine you're younger than you are.

For the professionals listening, one distinction

For the professionals listening, one distinction matters a lot. Most of these selfie checks run in what's called estimation mode. That means the system assigns an age to a face with no idea who the person is. Verification mode is different. It checks a face against a claimed identity, usually a document. Those are two separate tools with separate error profiles. Treat an age estimate like an identity match, and you've built a case on a guess. In plain terms, a system saying "probably an adult" isn't the same as a system saying "this is who you are."

Australia is testing a third path. According to Al Jazeera, Snapchat there can use a service called ConnectID. It leans on the identity your bank has already confirmed. The app gets back a simple yes or no, is this person old enough? No photo. No document upload. And the stakes for platforms are real. Companies that don't comply with Australia's teen ban face fines in the tens of millions of dollars.

Then there's Discord's behavioral approach. It's the least visible option. You never hold up an I.D. or snap a selfie. But you also don't get to see what signals the system read about you, or why it decided what it did. Invisible feels easier. Is it actually safer?

Not everyone thinks any of this should happen. At least sixteen U.S. states have passed social media age laws. Many face First Amendment challenges in court. Critics, including the privacy company Mullvad, argue that tying identity to speech can chill what people say. They point to activists in countries where criticizing people in power is dangerous. For them, anonymity isn't a convenience. It's protection.