Meta Youth Harms Trial Opens In Oakland With 29 States Lined Up

Twenty-nine states filed. Four are going first. The Meta youth harms trial opened in federal court in Oakland on 19 August, with California, Colorado, Kentucky and New Jersey leading, and it is expected to run six to eight weeks.

The case was filed in 2023 as multidistrict litigation. The allegation is that Meta knowingly designed Instagram and Facebook to be addictive to children, that those design choices contributed to a youth mental health crisis, and that the company collected data from users under 13. Meta rejects all of it.

What the states in the Meta youth harms trial have to prove

The interesting part is mechanical. The states are not arguing that social media is bad for teenagers in the abstract. They are arguing that specific product decisions, infinite scroll, variable-reward notifications, autoplay and engagement-optimised ranking, were shipped with internal knowledge of what they did to young users.

That turns it into a documents case. Internal research, growth targets and the emails around them will carry more weight than any expert witness. Jurors do not need to understand a ranking model. They need to read what the people building it wrote to each other.

Why the exposure number is so large

Meta has itself said potential liability could exceed 1 trillion dollars. That is not a forecast of what it will pay. It is a statement of what the underlying legal theory could support if it succeeds and then scales across every remaining state and every affected user.

There is precedent pointing in an uncomfortable direction for the company. A New Mexico court ordered Meta to pay 567 million dollars over youth harms, on top of 375 million dollars in civil penalties from a jury verdict in March.

What to watch for next

Three things. Whether documents unsealed at trial are worse than what is already public. Whether a verdict either way pushes the remaining twenty-five states to settle or to press. And whether product teams elsewhere quietly soften engagement mechanics for accounts they believe belong to minors.

That last one is already visible. OpenAI announced a dedicated ChatGPT experience for 13 to 17 year olds in the same week this trial opened, with age prediction routing suspected minors into it automatically. Regulators have been circling the design of consumer AI products for a while, from formal model evaluation regimes to the scale questions raised when one assistant reaches a billion monthly users.

Whatever the verdict, this is the first time the design decisions behind an infinite feed get tested in front of a federal jury. Product teams across the industry will read the transcript.