The Attention Machine

← Laws & Regulation

Courts & Liability

For decades, tobacco companies were protected by the argument that they sold a legal product. Liability came when courts accepted that the product was designed to be addictive and the companies knew. That argument is now being made — and won — against social media platforms.

How to read this page. Tap any underlined word to see the precise legal term and a short definition. Expand any "Deeper" box for the evidence and contested points. The main text works on its own — you can skip both and still get the whole argument.

The legal theory

Section 230 protects platforms from liability for user-generated content. But design choices — the infinite scroll, algorithmic autoplay, push notifications calibrated to moments of maximum psychological vulnerability — are not content. They are engineering decisions. Courts are now finding that platforms can be held liable for a defective design, even where they cannot be held liable for harmful content.

The wave of lawsuits

In October 2023, more than 40 US states sued Meta over harm to children and teens. The lawsuits consolidated into multidistrict litigation (MDL 3047) in the Northern District of California — the largest coordinated legal action against social media platforms in history. By May 2026, 2,527 lawsuits are pending in the consolidated proceedings. In January 2025, a California judge ruled that Section 230 and the First Amendment do not shield platforms from failure-to-warn claims. In August 2024, the Third Circuit found that TikTok must face a lawsuit after its algorithm exposed a 10-year-old to a dangerous challenge resulting in her death.

March 2026: the first verdicts

Juries in New Mexico and California delivered landmark verdicts against Meta and Google. In New Mexico, a jury found Meta violated consumer protection laws by misleading users about the safety of its platforms — awarding $375 million in civil penalties. In California, a Los Angeles jury awarded $6 million to a plaintiff who became addicted to Instagram and YouTube as a child, finding both companies acted with "malice, fraud or oppression." Instagram was found 70% responsible; YouTube 30%.

The tobacco comparison

The tobacco analogy is instructive and limited. In the 1990s, tobacco companies were found liable for knowing their product was addictive, concealing that knowledge, engineering the product to increase addictiveness, and targeting young people. All four elements are present in the social media litigation: Haugen documents showing internal knowledge of harm, features engineered to maximise addictive engagement, and children as a documented target demographic.

Where it breaks down: cigarettes cause physical addiction through a well-understood chemical mechanism. The causal pathway from social media design to psychological harm is real but more complex, and individual variation is significant. The analogy illuminates the litigation strategy; it does not describe an identical mechanism.

How we know — why the design theory works where content claims fail +

For two decades, suits against platforms died on Section 230: the harm was traced to user content, and platforms aren't liable for that. The litigation now succeeding reframes the target. It does not ask the jury to find a post unlawful; it asks whether the company's own product decisions — engagement-maximising features, notification timing, recommendation of harmful material to minors — were negligent or defective, and whether the company knew. Content is the user's; design is the company's. That reframing is what the January 2025 failure-to-warn ruling and the March 2026 verdicts accepted.

What the verdicts do and don't establish. They are first-instance jury findings — powerful signals, and near-certain to be appealed, so they are not yet settled appellate law. And note the framing discipline: outside the courtroom this site avoids calling users "addicted", because that wording shifts responsibility onto the user and is contested science. Inside the litigation, "addiction" is the plaintiffs' legal claim and the language of the verdicts, so we report it as such — the term belongs to the case, not to our description of users.

← Laws & Regulation