I have three children: one adult son and two younger kids, including a teenager and a daughter who is not yet 13. The older one can make his own choices now. The younger two still move through apps and games faster than I can see what appears on every screen.
Roblox worries me most. Not because every interaction is dangerous, but because a child can run into a chat rule, an age check, or a request for a face or ID before a parent even knows there was a question.
So when Meta agreed to an approximately $18 billion settlement over children’s online safety, the money was not the first thing I wanted to understand.
Will these new age checks actually protect children—or will they require families to hand over more information first?
🧭 Quick glossary
The short answer
Meta’s agreement adds meaningful protections. In participating U.S. states and territories, users identified as under 18 will receive a combined two-hour daily limit across Facebook and Instagram, a midnight-to-6 a.m. block, muted notifications during school hours, additional parent controls, and tighter content and contact settings.
But those protections depend on Meta knowing who is a child and who is an adult. That is the uncomfortable trade: the better a platform gets at identifying age, the more carefully we should ask what signals it uses, what it keeps, and what happens when it is wrong.
The agreement does not say every adult must upload a driver’s license. Meta says it uses profile and behavioral clues to identify likely teens, even when an account lists an adult birthday. Other age-check systems can use a selfie or government ID. The method—and the data handling around it—matters as much as the label “child safety.”
What Meta actually agreed to change
The headline number needs context. Meta describes the agreement as approximately $18 billion, paid in annual installments over 10 years. About 30% is conditional on TikTok and YouTube adopting comparable protections and making matching payments.
The part parents will notice is not the payment. It is the product design.
Like counts are also hidden by default, and cosmetic-surgery and extreme-makeup filters are blocked. Direct messages, however, are excluded from the Time Limit, Night Mode, and School Mode restrictions. That exception deserves attention: a child may be unable to scroll Reels at 1 a.m. but could still receive a direct message.
The age check may begin before anyone asks for ID
Age assurance is not one button or one technology.
Meta says its systems can examine the context of an account: a listed birthday, references to school grades or birthday celebrations, and other profile patterns that suggest the person may be underage. Meta also says its visual analysis looks for general patterns suggesting that someone is underage; it says this process does not create a facial fingerprint or identify a person in a photo.
That is different from asking a user to hold up an ID. It is also different from estimating age from a live selfie.
This distinction matters because parents online often ask, “Will everyone have to upload a driver’s license?” The current agreement does not establish that universal requirement. It requires Meta to improve its ability to detect accounts belonging to children under 13 and to place 13-to-17-year-olds into protected experiences.
If the system gets an age wrong, however, the family still needs a correction path. A teenager misclassified as an adult may miss protections. An adult misclassified as a teenager may face restrictions and be asked for stronger proof.
The privacy question is not imaginary
The Federal Trade Commission recognizes the same tension. Its 2026 COPPA policy says certain services may collect information solely to determine age without first getting parental consent—but only if they follow conditions. Among them: do not use the information for another purpose, keep it no longer than necessary, delete it promptly, protect it, disclose what is collected, and use a method likely to be reasonably accurate.
The Meta settlement has another wrinkle. Reporting on the executed agreement found that settling state attorneys general agreed not to bring certain claims over children’s data used for Meta’s age-assurance model, provided the data stays within the agreement’s limits. The data cannot be used for advertising, marketing, or ordinary recommendation optimization.
Still, a practical question remains unanswered in public detail: exactly which behavioral signals will Meta retain for model training, and for how long?
An independent auditor will review compliance. That is better than relying only on a company promise. It is not the same as giving parents a simple screen that says, “Here is what we used to estimate your child’s age, and here is when we deleted it.”
What should happen to age-check data
The intended path should be short and purpose-limited.
Not the purpose: ads, marketing, or ordinary recommendation optimization.
The protective purpose is only half the test. The verification information should have a short, limited life.
But my children use Roblox more than Instagram
Facebook and Instagram
Governed here by: Meta’s multistate settlement.
Main change: automatic teen protections, time and notification controls, stronger age assurance, and additional parent visibility.
Roblox
Not governed by: the Meta settlement.
Separate policy: a facial age check is required for users who want chat access where chat is available.
That boundary is easy to miss because families experience “the internet” as one moving stream. A child closes Instagram, opens Roblox, joins a game, and starts a chat. The legal rules and safety systems change while the device stays in the same hands.
In January 2026, Roblox began requiring a facial age check for users of all ages who want access to chat where the feature is available. Roblox says the images and video used for facial age estimation are deleted immediately after processing. It groups users by age for chat and limits communication between adults and children younger than 16.
Roblox also acknowledges that no system is perfect. Users can appeal an incorrect age result through alternatives that may include ID verification or a parent updating the child’s age through parental controls.
That is exactly the kind of moment I do not want my child handling alone—not because the prompt is automatically fraudulent, but because a face, an ID, and a birth date should never become routine taps.
Parent prompt decoder: what did the screen ask for?
Open the closest match. This is not an official Meta or Roblox checker, and it cannot determine whether a particular screen is legitimate.
“Confirm your birthday”
Pause before changing the date. Check whether the account birthday is accurate and whether the request appears inside the official app or website. A false adult birthday may remove protections; a false younger birthday may trigger restrictions.
“Take a selfie or video for an age check”
Confirm which company processes the image, what it says will be retained, and whether an alternative is available. On Roblox, facial age checking is tied to chat access. The Meta settlement itself does not say every user must complete a selfie check.
“Upload an ID”
Do not let a child proceed alone. Confirm the exact domain or app, why the ID is required, what fields can be covered, how long the image is retained, and how to appeal without an ID. Never send an ID through chat or email because someone claims to be support.
“Ask a parent to approve”
Use the parent’s own linked account or official parental-control page. Do not approve from an unsolicited message. Read what the approval changes—chat access, contacts, spending, privacy, or time limits are different permissions.
“Your age could not be confirmed”
Look for an official appeal path. Do not keep submitting more personal information to repeated pop-ups. Take a screenshot that excludes account IDs or other private details, close the prompt, and navigate to the platform’s official help center separately.
How it works: this guide matches a prompt to questions a parent can ask before sharing information. It does not inspect, upload, or store anything, and it cannot certify that a prompt is genuine.
The three rules I want at home
- Do not complete an age, face, or ID check alone. Bring the device to a parent first.
- A strange contact is not a reason to hide the screen. Showing it early should lead to help, not automatic punishment.
- Check each platform separately. Instagram protections do not carry into Roblox, YouTube, Discord, or the next app a child downloads.
This does not require reading every message or standing over every game. It creates a clear moment when a child knows, “This screen is no longer just mine to answer.”
What I will watch next
The settlement is more than a press release. The judge approved it, most terms must remain for 10 years, and an independent auditor will review Meta’s compliance.
Yet success should not be measured only by how many accounts Meta classifies. Parents need to know whether fewer unwanted contacts reach teens, whether age mistakes are corrected quickly, whether data collected for safety stays out of advertising systems, and whether children understand what to do when a verification screen appears.
I want platforms to recognize when a child needs protection. I also want them to know as little about that child as the job requires.
Those goals do not have to be enemies. But parents should not assume a shield icon proves both.
Sources
- Pennsylvania Office of Attorney General, settlement announcement and product requirements, August 26, 2026.
- Meta, agreement with state attorneys general and implementation details, updated August 27, 2026.
- Meta, how its AI-powered age assurance examines account context, updated July 31, 2026.
- Federal Trade Commission, COPPA age-verification policy statement, February 2026.
- Roblox, facial age checks required for chat, January 7, 2026.
- TechCrunch, reporting on the settlement’s children’s-data provision, August 27, 2026.
Leave a Reply