Meta's $18B Settlement Puts AI Age Verification to the Test

Meta faces a critical test of its artificial intelligence systems following an $18 billion settlement with state attorneys general over allegations the company harmed children's mental health through deliberately addictive platform design.

The agreement, involving 48 states plus the District of Columbia and U.S. territories, centers on a fundamental challenge: no child safety measure works if platforms cannot reliably identify which users are minors. Meta now has one year to demonstrate its age-checking technology can meet detailed requirements that the company suggests could become industry standards.

Why it matters

Age verification sits at the intersection of child safety, user privacy, and platform liability. As regulators worldwide intensify scrutiny of tech companies' impact on young users, the methods Meta develops under court oversight could define how the entire industry—including Google, TikTok, and Roblox—approaches this problem. The outcome will determine whether AI-based estimation can satisfy legal requirements or whether more invasive identity checks become unavoidable.

How AI age estimation works

Meta's current system analyzes multiple signals to estimate user ages rather than verify exact birth dates. On Instagram, the AI examines contextual clues including birthday celebration posts, school-related content, engagement patterns, friend networks, and login times. Someone consistently offline during school hours likely falls within school age ranges.

The technology also processes visual information from posted photos, analyzing physical characteristics like height and bone structure to estimate general age without identifying specific individuals. Meta combines these visual insights with text analysis and interaction patterns to flag potentially underage accounts.

Google employs similar AI-based age assurance for YouTube, differentiating adults from minors through watch history analysis. Gaming platform Roblox requires video selfies for age estimation when users want chat features, with the company claiming accuracy within one to two years for people aged 5 to 25.

Settlement requirements and limitations

Under the agreement, Meta must strengthen age-checking using both proprietary and third-party tools, subject to regular external audits. The settlement specifies targets for false positive rates—instances where minors are incorrectly classified as adults. When users identified as under 13 are removed from platforms, Meta must also verify the ages of their connections.

While Meta accepts government-issued IDs and promises deletion within 30 days, most users resist sharing driver's licenses or passports with tech companies. This reluctance drives the push toward AI estimation, though the technology cannot truly verify birth dates.

Some users and experts have raised concerns about accuracy, particularly for women and certain racial or ethnic groups. Identity verification firms like Yoti and Persona work with platforms including Meta and Roblox to process video selfies for age estimation.

The privacy dilemma

Meta has argued that app store operators like Apple and Google should handle age verification before users download apps. Those companies counter that such approaches fail to cover desktop computers, shared family devices, and pre-installed applications.

Digital rights advocates warn that age verification systems, regardless of method, require users to link offline identities to online activity through sensitive personal data vulnerable to leaks, hacks, or misuse. The Electronic Frontier Foundation cautions that determining what content harms minors involves subjective judgments that could expand age-gating requirements across the internet, from streaming services to local news sites, with First Amendment implications.

These details were first reported by Fortune.