Meta’s $18B deal with 29 states leaves room to use some children’s data for age-detection training

Meta’s recent settlement with 29 U.S. states, totaling $18 billion, is being read by privacy watchers as more than a headline number. According to a report by Yahoo Finance, the agreement contains a legal allowance that would let Meta keep certain data from children under 13 for specific purposes tied to age-detection technology, including training and testing models that estimate a user’s age.

The practical effect, as described in the report, is that the settlement does not simply end data-related disputes. Instead, it draws a boundary around what Meta may do with children’s data, while still permitting the kind of measurement and modeling required for age-gating systems. These systems are designed to determine whether a user is a child under 13, a threshold that triggers additional restrictions under U.S. children’s online privacy rules.

Age-detection models are not a static feature, and their performance depends on continual evaluation. In the report’s framing, the settlement effectively authorizes Meta to use certain retained data from under-13 users to support that evaluation. That means privacy compliance and product development are linked inside the settlement architecture, rather than treated as separate tracks.

The $18 billion figure and the number of participating states are central to how the deal is being characterized. With a settlement of this size, regulators and attorneys typically seek enforceable commitments, including limits on data handling and specific operational controls. In this case, the reported “pass” is notable because it suggests a negotiated carve-out for the technical realities of model development.

Meta did not lay out the full internal rationale for the term in the materials provided for this story. The Yahoo Finance report indicates the allowance exists, but it does not, in the information available here, spell out the exact categories of data permitted, the retention duration, or how the company would ensure that such data is used only for age estimation and not for broader targeting.

Meta, meanwhile, has a long-running business interest in age verification and related safety enforcement. Platforms that rely on automated systems to manage access and ad eligibility often need models that can reliably distinguish age groups at scale. For Meta, age-detection is part of a broader toolkit of safety and compliance measures that can affect user experience and downstream product and advertising operations.

It remains unclear, based on the information available here, how the settlement term interacts with other privacy requirements or whether additional safeguards, audits, or reporting obligations apply to the under-13 data retention for model purposes. Without the settlement text itself in the materials provided, the scope of the authorization and the conditions of use are still a key question.

For the months ahead, the biggest thing to watch is whether Meta or the states provide further clarification on what data categories are covered and what limits are placed around training and testing. Those details will determine whether the settlement’s “pass” is narrow and tightly controlled, or broader in practice, and how regulators measure compliance over time.