YouTube to use AI to estimate age to detect users under 18

YouTube will deploy an age estimation model to detect if U.S.-based users under 18 are using its video sharing platform. The roll out will begin from August 13.

Based on the estimation, the platform will provide a safer user experience for minors.

YouTube emphasised that it had followed this approach in other markets ahead of gradual rollout in the U.S.

The age estimation model uses signals such as YouTube activity and the longevity of the account to detect whether the user is a minor. If it deciphers the user to be under 18, it will notify them so they can submit a photo-based ID and verify their age.

In case YouTube flags the user as under 18, it will enable the standard protections for teen accounts on YouTube. These include showing only non-personalised ads, enabling “take a break” and bedtime reminders, showing reminders about privacy when making public actions, and minimising video recommendations that could be “problematic.”

While many children’s safety advocates are in favour of age verification to keep youngsters shielded from harmful content online, critics of the move have raised concerns about the privacy of adults being eroded, increased censorship of media, and the security of the personal ID data being collected by websites.

“These protections for teens are not new – we now have enhanced technology to more accurately determine whether or not a user is under 18 and are now able to extend these protections to more teenagers. We’ve used this approach in other markets for some time, where it is working well and we are now gradually rolling it out to the US,” said a Google employee in a YouTube post this week.