Skip to main content

Meta is expanding its use of AI face scanning to verify users’ age on Facebook Dating

Meta is expanding its use of AI face scanning to verify users’ age on Facebook Dating


Meta has used AI face scanning tools to verify the age of Instagram users. Now, the company says it will test the tool on Facebook Dating users.

Share this story

The Facebook logo on a blue background, surrounded by dark blue circles of various sizes
Illustration by Nick Barclay / The Verge

Facebook is testing ways for people to verify their age when using the platform’s dating app, including using an AI face scanning tool.

In a blog post today, Meta announced it would start prompting users on Facebook Dating to verify that they’re over 18 if the platform suspects a user is underage (you must be over 18 to use the dating service). Users can then confirm their age either by submitting a copy of their ID or by uploading a selfie video, which Facebook shares a portion of with a third-party company. Meta says the company, Yoti, uses facial features to estimate a user’s age without identifying them.

Meta says the new age verification systems will help stop children from accessing features meant for adults. It doesn’t appear there are any requirements for adults to verify their age on Facebook Dating (like making sure a 45-year-old isn’t pretending to be 18).

Meta has used Yoti for other age verification purposes, including vetting Instagram users who attempt to change their birthdate to make them 18 or older. Yoti says its system is broadly very accurate: it said that its rate of correctly identifying 13 to 17-year-olds as under 23 years old is 99.65 percent. Meta says age verification has sorted “hundreds of thousands” of people into age-appropriate versions of the app and that 81 percent of users prompted by Instagram to verify their age opt to use Yoti’s selfie option.

But the system isn’t equally accurate for all people: Yoti’s data shows that its accuracy is worse for “female” faces and people with darker complexions. Researchers have flagged concerns over how facial recognitions and analysis software perform differently for people along age, race, and gender lines.