How a $6M Verdict Led to Meta's $18B Settlement, and What It Means for Consumer Chatbots

Meta settled with 48 state attorneys general on Wednesday morning, eight days into a federal trial in Oakland over whether it designed Instagram and Facebook to addict children. Instagram head Adam Mosseri was in the middle of his testimony. Mark Zuckerberg was scheduled to take the stand next.

Before this year, no jury had ever found a social media company liable for designing an addictive product. This is the fourth time Meta has been on the hook since March. Meta pays when the executives get close to the stand. The two times it let a jury decide instead, it lost.

The scoreboard, in order:

• New Mexico, March 24. A Santa Fe jury found 75,000 violations of the state's Unfair Practices Act and imposed the statutory maximum on each, $375 million. In August, after a separate bench trial, the judge added a $567 million abatement fund. Total: $942 million.
• KGM v. Meta, March 25. A Los Angeles jury returned $6 million, $4.2 million of it against Meta. Zuckerberg had spent nearly eight hours on the stand five weeks earlier.
• Breathitt County, Kentucky, May. The first test case in the federal litigation. Settled on the eve of trial, roughly $27 million for one rural school district.
• Oakland, this morning. Roughly $18 billion across 10 annual installments.

Two of the four reached a jury. Meta lost both, on consecutive days. It settled the other two before a jury could rule.

They Sued Over the Product, Not the Posts

Every one of those outcomes traces back to a choice made before a single witness was sworn.

The complaints were not about content. They were about design. Infinite scroll, autoplay, notification timing, age gates.

That distinction is the entire ballgame. For nearly 30 years, Section 230, the 1996 statute that shields platforms from liability for what their users post, swallowed cases that asked a court to punish a company for what appeared on its screens. Recast the same conduct as a defectively designed product and the claim suddenly has somewhere to stand, which is what a Los Angeles judge confirmed when he denied Meta's post-trial motions on Section 230, the First Amendment, and causation in a single order.

The proof, once they got there, came out of Meta's own filing cabinets.

An internal document told the jury that if the company wanted to win big with teens it had to bring them in as tweens. Another showed that eleven-year-olds returned to Instagram at four times the rate of competing apps, on a platform that officially required users to be thirteen. Nobody had to invent new science. They had to survive long enough to read the mail.

Then one small verdict set the price for everything above it. Six million in March. Twenty-seven million in May. Eighteen billion in August.

Nobody negotiated their way to that last number. The twelve jurors who returned the first one did. Ten of them, anyway, with two voting for the defense on every question on the form.

The Chatbot Makers Are Next

Four years. That is the distance between a theory no court had accepted and what North Carolina Attorney General Jeff Jackson called the largest settlement any state has reached with a technology company.

For most of those four years, the firms carrying these cases were losing motions, funding inventory against a verdict that did not exist, and getting no press for any of it.

The lawyers collecting today are the ones who filed when the theory was still a guess.

They are roughly where Meta was in 2022.

No adverse verdict anywhere. Confident First Amendment defenses and an untested argument that Section 230 covers what a model itself says. Product decisions documented internally by engineers who wrote them down, because engineers always write them down.

The clock is running faster this time. Garcia v. Character Technologies was filed in the Middle District of Florida in October 2024 by the Social Media Victims Law Center, the same firm that represented the plaintiff in KGM. It survived dismissal seven months later and settled in January. Raine v. OpenAI, filed the following August over a 16-year-old's suicide, pleads design defect and failure to warn and names Sam Altman personally.

Meta agreed on Wednesday to cap teen usage at two hours a day, shut its apps down overnight, and check how old its users actually are. It spent the last four years insisting none of that was necessary.

Someone is going to extract the same concessions from one of them. It will not take four years.

Adam Ramirez, Managing Editor of The Tort Report, previously edited and covered law and business at Bloomberg Law, Forbes, and Thomson Reuters.