After reaching a settlement with the attorneys general of 29 states in the United States, in addition to paying up to $18 billion and implementing additional child safety measures, the agreement also included a less common provision. According to foreign media, the states agreed not to pursue legal action regarding Meta's retention and use of relevant data in training and testing models for identifying children's ages, and will continue to prosecute in accordance with existing child privacy laws.

The settlement terms allow for limited exemptions.

According to the agreement, Meta is required to develop, train, and begin testing a model for identifying users under the age of 13 within one year after the agreement takes effect. The document does not explicitly state that AI must be used, but Meta's existing age identification tool already relies on AI technology.

It is also stipulated in the agreement that this exemption is limited to the training and testing of age recognition models. Meta Data of users under the age of 13 shall not be used for ad targeting, marketing, or algorithm optimization.

COPPA Law Enforcement Space Comes Under Attention

The U.S. Children's Online Privacy Act typically requires websites and applications to limit the collection and storage of children's personal information. Foreign media pointed out that this settlement agreement not only attempts to leave room for operations regarding the training of Meta's models but also includes stronger disclaimers, stating that the prosecutors of the relevant states "completely, finally, and permanently" waive any past, present, or future claims under COPPA and similar state laws regarding the use of such data.

Legal professionals believe that it is not entirely unusual for Meta to seek such protection, as companies may indeed need to retain a small amount of data for compliance verification when dealing with deletion requests and identity verification scenarios. However, the issue lies in the fact that COPPA falls under federal law, and the primary enforcement agency is the United States Federal Trade Commission (FTC), whereas FTC is not a signatory to this settlement agreement.

Data isolation and supervised execution are more complex.

Foreign media believes that the real challenge lies at the implementation level. Meta requires isolating child behavior signals and related data from other systems of the company, and limiting their use to identifying and removing accounts of individuals under the age of 13. However, the data and models within large platforms are often interrelated, making it difficult to achieve complete technical and organizational isolation over the long term.

The agreement states that an independent auditor will participate in supervising whether Meta complies with the settlement terms. However, it is still unclear to the outside world which data Meta will retain for model training, how much behavioral information is involved, and for how long it will be stored.

Several lawyers stated that even if Meta uses this data beyond the agreed scope, the state attorney general can still hold them accountable. However, in the event of future disputes, the focus of the case is likely to shift to whether the use of such data remains within the limits permitted by the settlement agreement. This also means that subsequent law enforcement and evidence presentation may become more complicated.

The dispute extends to the issue of data usage regarding AI.

This controversy also touches on a broader issue in the AI industry: if models are to improve their recognition and judgment capabilities, they often require more user data support. For Meta, identifying which accounts belong to children may also require a more in-depth analysis of the usage behavior of underage users on social platforms.

Foreign media believes that this further highlights the tension between child protection and model training. The settlement agreement attempts to draw an enforceable line between the two, but whether this line can be clearly monitored in the future remains a focal point of external concern.