Let me tell you what most age verification on the internet actually is right now.
You open your phone. A camera points at your face. An AI guesses how old you are. If it thinks you look old enough, you get in.
That is Age Estimation. And it is the dominant solution being deployed right now by platforms scrambling to comply with new age assurance laws.
Here is the problem. It is a guess. Not a calculation. Not a proof. A probability. A model trained on images, making an inference about your face, with a margin of error of up to 8 years. The Australian Government’s Age Assurance Technology Trial tested around 50 vendors across every method available. Age Estimation showed significant demographic bias and variance. Document-based verification ranked highest for accuracy and privacy. It wasn’t close.
So why is everyone using it?
Because it is frictionless. No app. No document. Just a selfie. Platforms love it because users don’t drop off. Regulators tolerate it because something is better than nothing.
But that window is closing fast.
Australia’s SMMA is live. Texas SB2420 just kicked in. The EU’s EDPB has come out explicitly in favour of evidence-based verification over probabilistic methods. The direction is clear: a guess is not going to be good enough for much longer.
Three things that make Age Estimation dangerous
It can be spoofed. A deepfake, a photo, anything that looks like a face to a camera. No liveness binding, no document tie. Easy to beat.
It has no audit trail. No persistent credential, no verification token. When a regulator asks you to prove your compliance process worked, there is nothing to show.
And it collects biometric data. That face scan goes to a vendor’s server. You are adding to a surveillance layer every time you log in, even if nobody calls it that.
We built something different.
ShareRing verifies identity against a real document, once. After that, you tap to share a cryptographic proof. The merchant gets a yes or no. Your ID stays on your device. Nothing is stored centrally. Nothing is shared that doesn’t need to be.
That is not the easiest thing to build. It is the right thing to build.
We just relaunched our Compare Us page to lay all of this out side by side, no spin. Every approach, every tradeoff, including Age Estimation, scored honestly.
If you are building something that needs to verify age, or evaluating what your current provider is actually doing, it is worth a look.
