Social media and technology companies are increasing their efforts to identify which of their users are children, as pressure from governments, regulators and the public grows over online safety.

The issue has become a key part of Meta’s recently announced $18 billion legal settlement over allegations that its Facebook and Instagram platforms harmed children’s mental health by deliberately designing features that kept young users engaged.

The settlement, involving 48 US states, the District of Columbia and US territories, places strong emphasis on improving age assurance — the process of determining whether a user is a child or an adult.

The challenge is simple but difficult: safety measures designed for children are of limited use if online platforms cannot accurately identify which users are underage.

Other major technology companies, including Google, TikTok and Roblox, are also strengthening their systems for identifying younger users.

Under the settlement, Meta will have one year to improve its age-checking system using its own technology as well as third-party tools. The company will also face regular independent audits to assess how effectively the system works.

AI increasingly used to estimate users’ ages
In the physical world, people often use government-issued identification cards to prove their age. Online, however, many users are uncomfortable sharing documents such as passports or driving licences with large technology companies.

Meta currently accepts identification documents for age checks and says it deletes them within 30 days. But companies are increasingly turning to artificial intelligence to estimate a user’s age.

Unlike traditional age verification, AI cannot confirm a person’s actual date of birth. Instead, it examines different clues to estimate whether someone is likely to be a child or an adult.

Google began using an AI-based age assurance system for YouTube last year, analysing users’ viewing behaviour to help distinguish adults from minors. Meta also uses AI to identify potentially underage accounts.

On Instagram, the technology can analyse clues such as posts about birthdays or school, the type of content users like or comment on, their online activity, friendships and the times of day they use the platform.

Meta has also said its AI can examine photos for general visual clues, such as physical features that may help estimate a person’s age. The company says it does not use the system to identify the specific person in an image.

Under the settlement, Meta has also agreed to set targets to reduce cases where minors are incorrectly identified as adults. If an account is found to belong to a child under 13 and is removed, Meta will also review the ages of the user’s friends.

Face scans also being used
Some technology companies are using video selfies and facial analysis to estimate users’ ages.

Identity verification firms such as Yoti and Persona provide such technology to companies including Meta and Roblox.

Roblox, which allows children under 13 to use its platform, asks users to submit a video selfie if they want to access certain chat features. The company says the videos are deleted after the age-checking process is completed.

However, facial age estimation has raised concerns over accuracy. Some users have reported incorrect results, while experts have questioned whether the technology works equally well across different groups, including women and people from different racial or ethnic backgrounds.

Roblox has said its system can generally estimate the ages of users between about 5 and 25 within one or two years.

Debate continues over who should verify users’ ages
Meta and some other companies have argued that app store operators, such as Apple and Google, should be responsible for verifying users’ ages before allowing them to download apps.

Apple and Google have opposed the idea, arguing that such a system could fail to cover people using desktop computers, shared family devices or apps already installed on a device.

Privacy remains a major concern
While age checks are intended to protect children from harmful or inappropriate online content, critics warn that widespread age verification could threaten users’ privacy and anonymity.

Digital rights groups have raised concerns that people may eventually have to prove their age to access a wide range of online services, including news websites, health information, entertainment platforms and other forms of online content.

Critics also argue that collecting identity documents, facial scans or other personal information could create security risks if such data is leaked, hacked or misused.

The debate highlights a growing challenge for technology companies and governments: how to better protect children online without creating a system that requires everyone to give up large amounts of personal information to use the internet.