Google will soon use machine learning to estimate the age of its users. In an update on Wednesday, Google said it’s testing a machine learning model in the US to help determine whether someone is under 18, allowing it to “provide more age-appropriate experiences” across its platforms.

YouTube CEO Neal Mohan first revealed the change in an annual letter published on Tuesday. The age estimation model will use existing data about users, including the sites they visit, what kinds of videos they watch on YouTube, and how long they’ve had an account to determine their age. When it detects a user may be under 18, Google will notify them that it has changed some of their settings and will offer information about how users can verify their age with a selfie, credit card, or government ID.

Google will apply its existing safety features to underage accounts, including its SafeSearch filter that helps remove explicit content from search results. It will also restrict content on YouTube that might not be appropriate for users under 18. Google plans to bring its age estimation technology to “more countries over time.”

“We will continue to explore additional opportunities to provide increased transparency to users about age estimation at the account level,” Google spokesperson Matt Bryant said in a statement to The Verge.

In the next week, Google will let parents limit calls and message notifications on their child’s phone and tablet during school, something it first announced last year. Starting next month, parents will be able to add contacts to their child’s phone through the Family Link app, limiting phone calls and texts to only those numbers. Additionally, Google says it will soon allow parents to add and manage payment cards in their child’s Google Wallet.