SAN FRANCISCO, 11 AUG 2026 — Meta, TikTok, Snap and Google tried to end thousands of addiction lawsuits before trial by invoking Section 230. On Monday the Ninth Circuit told them the argument would have to wait.
The appeal was denied on a precise ground worth quoting properly: Section 230 provides a defence to liability, not immunity from being sued. Because it is a defence, it is something you raise at trial, which made the appeal premature. More than 3,000 consolidated federal lawsuits now proceed.
What the ruling did and did not decide
The distinction is subtle, but it is the entire basis for the ruling.
The platforms argued that Section 230 of the Communications Decency Act of 1996 shields them from claims that they failed to warn users about addictive design. The court did not rule that the shield fails. It ruled that the shield is the wrong shape for what they wanted to do with it — a defence you argue, rather than a gate that keeps you out of court in the first place.
That distinction has been available in the statute all along and it has rarely mattered so much, because for most of Section 230's life defendants have been able to get cases dismissed early on its strength. Being told the argument survives but the early exit does not is a meaningful narrowing of its practical value.
A defence that only works after a trial is a much weaker asset than one that ends a case at the pleadings; the cost, discovery and disclosure of a trial are most of what a defendant is trying to avoid.
What the cases actually allege
Thousands of suits have been consolidated into a single federal action, brought by private individuals, state and local governments, and school districts.
The core allegation is that these companies intentionally designed their products to be addictive to minors. This framing is a deliberate legal tactic. By claiming this is about product design rather than user content, plaintiffs hope to sidestep the Section 230 shield entirely.
If a case is about what somebody wrote, Section 230 is a strong defence. If it is about how a recommendation system, a notification schedule or an autoplay behaviour was built, the plaintiffs are arguing it is a product-liability case that happens to involve a website. The Ninth Circuit has left it for a trial to decide whether that distinction holds, and the entire litigation turns on the answer.
A separate trial starts on Wednesday
In the same decision the court declined Meta's request to postpone a trial beginning on Wednesday, brought by 29 state attorneys general.
That case alleges Meta illegally collected and used children's data, designed its platforms to keep young users engaged, and misled consumers about their safety. It is a different action from the consolidated private litigation, and it is the one that will produce evidence in public first.
Anyone deciding how much attention to give this should give it to Wednesday rather than to Monday's procedural ruling. Discovery in a case brought by 29 states, argued in open court, is where the internal documents get read out.
Two juries have already found against Meta
Meta has lost two lawsuits on similar issues, reported as the first time the company has been held liable over child safety in jury trials.
That matters more than an interim procedural ruling, because it establishes that a jury presented with this theory can accept it. Until a plaintiff wins, a novel legal theory is a hypothesis; afterwards it is a settlement risk with a number attached.
It also explains the urgency of the appeal. A defendant facing thousands of consolidated claims wants a ruling that disposes of them as a class. Forcing them to trial confirms that their exposure is real and will be established case by case.
Why this reaches beyond the four companies
The design-not-content framing, if it survives, would apply to any system optimised to hold attention — a category much broader than social networks.
Recommendation systems, engagement-optimised feeds and habit-forming notification patterns are now standard components of consumer software generally, including products aimed at children that nobody thinks of as social media. Such a finding would make designing for compulsive use an actionable claim against any company that builds such software.
Which is the reason to watch this from outside the United States too. Regulators elsewhere have approached the same concern through codes of practice and age-assurance rules; American courts are approaching it through product liability, and product liability produces damages figures that change corporate behaviour faster than compliance obligations do.
What to watch
Whether the platforms seek further review, and whether any court takes the Section 230 question on the merits before a trial happens.
Whether the consolidated case reaches trial and what the first verdicts look like at that scale. Two individual losses are a signal; a consolidated outcome is a number.
And whether any settlement emerges. Thousands of claims backed by governments and school districts, with two adverse jury verdicts behind them, is the shape of a case that settles rather than concludes.