Meta is expanding its AI-powered age estimation and verification system, described in detail in May, globally on Instagram, and will then bring it to Facebook and Threads.
This is not the only place Meta is using this technology. “We’re continuing to roll this technology out on Facebook and, later this year, we’ll begin launching similar technology globally on Threads,” the announcement said.
The system, that the giant says is able to find people it suspects are teenagers “even if they use an adult birthday” and put them into Teen Account restrictions, is based on machine learning algorithms that analyze profiles and “contextual clues” to determine age.
This means that even if you are an adult who has been using Instagram without restrictions, but the giant’s new system decides that you are likely a teen, you will be treated as one and have your account restricted.
Meta also says that it uses this system “to remove people under 13 from our services” – while those who are put in the Teen Account category have limits imposed on who can contact them, what content they can see, and their time spent on the apps.
Parents in the US will this month start seeing an in-feed video on Instagram explaining the way the system works.
As for how it works – Meta previously explained that it uses “visual analysis” that scans photos and videos “for visual clues about a person’s age” – but, the company insists, “this is not facial recognition.” The AI also analyzes the content of posts, comments, and bios to look for “contextual clues” – such as birthday celebrations or school grades.
If the AI suspects that someone is lying about their age, for example by changing their birthday from under 18 to over 18, “we require them to verify their age using an ID or Yoti’s facial age estimation tools,” Meta said. In other words, an adult who is misread by the system has to provide ID or a face scan to prove their age.
In Australia, Meta is using Apple’s age-range signal, that the tech giant provides to apps at each download, as another tool to determine age.
Meta said it is using this signal “voluntarily” – but would very much like to make it mandatory for all app stores and developers. “We’ve long supported legislation requiring age verification and obtaining parental consent on the app store,” the company said.
Meta argues that getting age information from the device is “far more effective and privacy-preserving than asking every platform to verify age individually”.
Meta is not pleased that some legislation “require invasive age checks for everyone” – presenting its own system as the standard.
Meta is tying the announcement of the global expansion of its system to an agreement it reached with 52 bipartisan attorneys general across US states, territories, and the District of Columbia.
Under the settlement, which was announced in August, Meta agreed to implement a number of measures, including “investing in even stronger technology to proactively catch accounts that may belong to children under 13” and strengthening “the technology we use to identify accounts that may belong to teens between the ages of 13 and 17.”
The agreement also requires Meta to use “age assurance” for every user in the settling states, and names “ID verification and facial age estimation” as two of the methods. Another is to use Apple and Google’s device age signals. And if a user’s age is not verified after 14 days, they are treated “as a Teen User regardless of stated age.”
Meta is now calling on other Big Tech companies – TikTok and YouTube – to “adopt these same practices to keep teens safe.” The August settlement actually requires Meta to make sure these companies, as well as Snap, adopt the same practices.
What this shows is that the industry is building the infrastructure of the digital ID gate – and is asking governments to make it the law of the land.
