Meta’s New AI Can Catch Teenagers Lying About Their Age. Here’s How.

Meta is expanding its artificial intelligence-powered age-estimation technology globally, allowing Instagram to identify users it suspects are teenagers even when they have entered an adult date of birth. The announcement, made on November 9, marks another step in the platform’s attempt to move beyond self-declared age verification and proactively identify underage users, globally. 

The company says the expansion will place millions more teenagers into its Teen Account protections. These accounts come with restrictions on who can contact users, the kind of content they can access and how their profiles are configured. Accounts are private by default, messages from strangers are restricted, and certain sensitive content is filtered out. The technology is particularly interesting because Meta is not relying on users to tell it how old they are. Instead, it is attempting to infer their age from their digital footprint.

So How Does It Work?

Meta’s AI examines entire profiles for contextual clues that could indicate someone is a teenager. These include references to school grades, birthday celebrations and other age-related information appearing in posts, comments, captions and profile biographies. The company has also described using visual analysis of photographs and videos to identify general age-related patterns, including physical cues such as height and bone structure.

Meta has clarified that its visual analysis is not facial recognition. It does not create facial fingerprints or attempt to identify the individual in an image. Instead, it looks for visual characteristics that could suggest that someone is underage.

The company also uses activity-based signals and other age-assurance methods. However, it has not disclosed the complete list of signals, their relative importance or the algorithm’s accuracy across different demographic groups.

There is an important distinction here. This particular global expansion is about identifying users between 13 and 17 and placing them in Teen Account protections. Detecting and removing children under 13 is a separate enforcement process.

The move also comes amid growing regulatory and legal pressure on social media companies over their impact on young users. Meta has faced lawsuits alleging that its platforms contribute to harm among children and teenagers, including addiction. Its November announcement also builds on commitments made under an agreement with 52 US state attorneys general to strengthen the identification of underage users.

Different Approaches

Interestingly, Meta is also experimenting with a different approach in Australia. It has started using Apple’s age-range signal as an additional indicator. When parents configure a teenager’s device, Apple’s system can assign an age range and share that information with app developers when the user downloads an application. Meta argues that a common age-assurance mechanism at the app-store level could be more consistent and privacy-preserving than requiring every platform to independently establish a user’s age.

But this raises another set of questions.

Estimation Vs Verification?

For starters, AI-based age estimation is not the same as age verification. An algorithm can identify patterns suggesting that someone is 16, but it cannot establish their actual date of birth with certainty. What happens when an adult is incorrectly classified as a teenager? Conversely, how many teenagers will successfully evade detection?

There are also privacy concerns. Even if Meta is not using facial recognition, analysing photographs, conversations, interactions and behavioural patterns to infer personal characteristics raises questions about how much information a platform should be allowed to process.

Algorithmic bias is another concern. Visual characteristics and behavioural patterns do not necessarily translate into accurate age estimates across different populations. Meta has yet to publish sufficiently detailed independent accuracy assessments that would allow users to evaluate these risks.

Does It Apply To India?

For India, the development deserves attention, particularly as debates around online child safety, age verification and platform accountability gain momentum. However, Meta has not announced a country-specific implementation schedule for this latest expansion, and it would be premature to assume that every Indian account is already being assessed through all these methods, because it is applicable globally.

Privacy Question? 

Ultimately, the challenge is not whether children should be protected online. That is a legitimate responsibility for platforms. The bigger question is whether the technology companies entrusted with that responsibility should also have power to infer a user’s age from their digital behaviour. So is this surveillance and a breach of privacy? What are the regulations associated with it? These are the unanswered questions.

Instagram may be trying to make its platform safer for teenagers. But in doing so, it is also moving towards an internet where platforms increasingly decide who users are, rather than simply accepting who they say they are.