Meta has agreed to an $18bn settlement that obliges the social-media giant to enhance its AI-powered age-assurance systems. The deal, signed with 48 US states, the District of Columbia and several territories, aims to curb the presence of under-age users on Instagram and Facebook.
Settlement details
The agreement resolves claims that Meta deliberately designed its platforms to hook young people, harming their mental health. Under the terms, Meta has a year to refine its approach, using both its own tools and third-party solutions, while subjecting the process to regular external audits. The settlement also sets targets for false-positive rates, cases where minors are mistakenly classified as adults, and requires the company to check the ages of friends of any account identified as under-13.
AI age-assurance technology
In the offline world, government-issued IDs prove age, but many users are reluctant to share such documents online. Meta currently accepts IDs and promises to delete them within 30 days, yet it is increasingly turning to artificial intelligence to estimate ages. The AI analyses contextual clues such as birthday posts, school-related content, interaction patterns and login times. Visual cues from photos, including height and bone structure, are also examined, though the system does not identify the individual in the image.
"By combining these visual insights with our analysis of text and interactions, we can significantly increase the number of underage accounts we identify and remove," Meta explained in a blog post.
Google has adopted a similar AI-driven system for YouTube, and other platforms such as Roblox use video selfies to estimate age, deleting the recordings after verification. While Roblox claims its method can estimate age within one or two years for users aged five to 25, critics warn of inaccuracies, especially for women and certain ethnic groups.
Privacy and regulatory challenges
Age checks aim to keep children away from unsuitable online spaces, yet privacy advocates argue that any system requiring users to hand over sensitive personal data creates new risks. The Electronic Frontier Foundation has long warned that "(No) matter the method, every system demands users hand over sensitive and immutable personal information that links their offline identity to their online activity," noting the potential for leaks, hacks or misuse.
Google has also criticised proposals that shift verification responsibility to app-store owners, stating that such measures would not cover desktop computers or shared family devices and could be ineffective against pre-installed apps.
"Billed as 'simple' by its backers, including Meta, this proposal fails to cover desktop computers or other devices that are commonly shared within families. It also could be ineffective against pre-installed apps," Google wrote.
European regulators are watching the development closely, as the settlement could influence future EU policy on digital safety and data protection. If Meta's AI tools prove effective and respect privacy safeguards, they may become a benchmark for the industry. Conversely, any missteps could trigger stricter oversight under the EU's Digital Services Act and the General Data Protection Regulation.
What comes next?
Meta now faces the task of delivering a robust, privacy-respecting age-verification solution within twelve months. Ongoing audits will assess the accuracy of the AI models and the rate of false positives. Meanwhile, privacy groups and lawmakers will likely scrutinise the rollout, potentially shaping how AI-based age checks are implemented across the European digital market.