Age assurance moves to center stage after Meta child safety settlement
When Meta agreed to pay $18 billion to resolve claims that it harmed young users, it also accepted something less discussed but more consequential: a set of age assurance commitments that are the most detailed of any child safety measure in the settlement. The reason is straightforward. Even the strongest safety controls are of limited use if the company cannot reliably identify which accounts belong to minors.
Age assurance is now the linchpin of child safety strategies across the wider industry. Google, TikTok, Roblox and others are racing to strengthen their own systems as public and regulatory scrutiny intensifies in the U.S. and overseas. Meta has said the protections it agreed to could become industry standards, and the settlement gives the company a year to put them in practice. But the timeline only adds to the pressure, because the technology is still evolving and the privacy stakes are high — not just for children, but for adults who may be forced to prove they are over 18.
The agreement, announced last week with the Menlo Park, California-based company, was reached with 48 states, the District of Columbia and U.S. territories. The state attorneys general alleged that Meta intentionally designed Instagram and Facebook to hold young users' attention in ways that harmed their mental health.
In the physical world, proving age usually means showing a government-issued ID. That works online in theory, but many people are uncomfortable letting Meta, Google or other tech giants see their driver's license or passport. Meta currently accepts IDs for age checks and says it deletes the records within 30 days. But ID checks are only part of the picture.
A growing number of companies are using artificial intelligence to estimate age. This is not the same as age verification. AI can often guess how old someone appears to be, but it cannot verify a person's actual birth date. Last year, Google introduced an age-assurance system for YouTube that uses AI to distinguish adults from minors based on watch history. Meta uses similar techniques.
On Instagram, Meta says its AI considers “contextual clues” such as birthday posts, school-related content, interactions, friend connections and login patterns. A user who is consistently offline during school hours, for example, is likely to be school-aged. Meta also says its AI can examine photos for physical clues, including “height or bone structure, to estimate someone's general age; it does not identify the specific person in the image.” The company says combining those visual signals with text and interaction analysis significantly increases the number of underage accounts it can find and remove.
Under the settlement, Meta is required to strengthen its age-checking tools using both its own systems and third-party services, with regular outside audits of how well the tools perform. The agreement sets explicit goals around false positives — cases in which minors are mistakenly identified as adults. It also requires Meta to check the ages of a removed user's friends if the user has been identified as under 13.
Companies like Yoti and Persona, which work with Meta and Roblox, rely on video selfies to estimate age. Roblox, which unlike some platforms allows children under 13 to sign up, asks players who want to chat with other users to submit a video selfie for an age estimate. The company says the video is deleted once the check is processed. The face scan is not required simply to use Roblox, only for chatting.
Some users have reported mistakes, and experts have cautioned that facial age estimation can be less reliable for women and for certain racial or ethnic groups. Roblox chief safety officer Matt Kaufman said last year that the system can estimate a person's age within a year or two for users between roughly 5 and 25.
Some platforms, including Meta, have argued that age verification should happen at the app store level, with companies like Apple and Google verifying a user's age before allowing app downloads. Apple and Google have rejected that idea. Google argued in a blog post last year that the proposal, “billed as 'simple' by its backers, including Meta, fails to cover desktop computers or other devices that are commonly shared within families. It also could be ineffective against pre-installed apps.”
Age checks are designed to keep children out of spaces that are unsafe or inappropriate, whether social media platforms or adult websites. But critics of age verification warn of a less secure, less private and less free internet. If the trend continues, people could be asked to verify their age before accessing not just adult content but also news, health information and the ability to speak openly and anonymously. What counts as harmful to minors is often subjective, and legal experts say broad age-verification mandates could run into First Amendment problems. The requirement could eventually extend to anything from Netflix to a neighborhood blog.
The Electronic Frontier Foundation has spent years warning about “age-gating the internet.” In one post, the digital rights group argues: “No matter the method, every system demands users hand over sensitive and immutable personal information that links their offline identity to their online activity. Once that valuable data is collected, it can easily be leaked, hacked, or misused.”
The next year will show whether Meta can meet its obligations without triggering a broader backlash. The outcome is likely to shape not just the company's apps, but the rules governing age checks across the entire internet.