Two verdicts in forty-eight hours. On March twenty-fourth, a court in New Mexico found Meta liable for failing to protect minors from sexual predators on its platforms, with a penalty of three hundred seventy-five million dollars. The following day, in Los Angeles, another jury found Meta and YouTube liable for an addictive design that damaged a young woman's mental health. That coincidence is not accidental. It represents a convergence in the courts, and what both rulings signal together is something the tech industry has skillfully dodged for years: direct responsibility for the way it builds its platforms.

What the juries evaluated was not user-generated content. That had been tried before, and Section 230 of the United States communications law systematically blocked those lawsuits for decades, shielding platforms from any blame for what third parties post. This time, the lawyers shifted their approach: they questioned the product's very architecture. The infinite scroll. The variable-reward algorithms that determine what to show, at what moment, and to whom, with the express purpose of prolonging screen time. The jury accepted that framing. That shift—from content to engineering—is generating legal repercussions we are still measuring.

The internal documents presented at trial are revealing. Not because they expose a bug or a miscalculation, but because they document a deliberate choice. One Meta executive wrote, verbatim: "If we wanna win big with teens, we must bring them in as tweens." Another record showed that eleven-year-olds returned to Instagram four times more often than to other apps, despite the platform requiring users to be at least thirteen to sign up. That is not a flaw. It is an explicit design goal. Researchers have argued for years that the distinction between error and intent defines negligence. Here, that distinction was laid out plainly in the company's own files.

When a model is optimized for a single metric, everything else becomes an externality. Uncounted costs end up falling on others. In this case, those others were teenagers. Meta did not create flawed platforms: it created platforms that operated exactly as their incentives dictated. The files also indicate the company kept "beauty" filters despite warnings from eighteen internal experts about the impact on young users' body image. The harm wasn't incidental. It was a cost that others bore.

The comparison to the tobacco industry goes beyond rhetoric. It's structural. Tobacco companies knew the risks, documented them internally, denied them publicly, and fought them in court for decades. The turning point in that industry didn't come from a single ruling, but from an accumulation of case law, global regulatory pressure, and, eventually, penalties that made the business model unsustainable. Today there are at least two thousand four hundred seven consolidated lawsuits in a federal court in California, including eight hundred from school boards attributing counseling and safety costs for their students to Meta, TikTok, and Snapchat. Another federal trial is scheduled for June twenty-twenty-six, where a judge could impose algorithmic changes, such as banning infinite scroll for minors. The dynamic repeats. The scale keeps growing.

The reach extends beyond the United States. In Australia, new age-restricted content codes took effect in March twenty-twenty-six, and legal analysts already see the verdict as a tool for local litigation. In the United Kingdom, annual social costs from mental illness exceed ninety-four billion pounds; if part of that is causally linked to platforms, the economic damages would dwarf every prior case. In the Philippines, the debate over a total ban for minors is gaining concrete judicial backing, beyond mere political argument. TikTok and Snap, which settled before trial, use mechanisms similar to those declared addictive. The settlement doesn't erase that classification.

What adds tension happened one day later. On March twenty-sixth, as the verdicts spread worldwide, the European Parliament approved its position on the AI law, exempting sectors like banking, healthcare, and infrastructure from stricter rules for their systems. Two simultaneous moves, in opposite directions, facing the same question: who answers when algorithmic design causes harm? In California, a jury said someone must answer. In Brussels, parliament indicated that already-regulated sectors don't need additional accountability under the new framework. That gap isn't just geographic. It's foundational. And it won't close on its own.

What these rulings don't resolve also deserves attention. Six million dollars for Meta and Google amounts to a minimal fraction against capital budgets exceeding one hundred billion dollars annually for each. A penalty that doesn't shift the internal cost-benefit balance doesn't change the design. It remains to be seen whether the punitive damages still to come, the accumulated regulations, and the two thousand four hundred pending cases achieve what the initial verdicts did not: realigning incentives. As long as screen time remains more profitable than user protection, priorities won't change. There are signs that wellbeing-centered designs could gain ground if judicial pressure persists, but that depends on the pressure not letting up.

Social media didn't invent the capture of attention. Casinos refined variable rewards long before infinite scroll existed. Advertising has optimized exposure time for over a century. What's new is the scale and precision: algorithms personalize the hook for each user, in real time, for billions of people. And they fine-tune it for the most vulnerable: young people with developing brains, more sensitive to reward stimuli. These dynamics tend to concentrate power rather than distribute it, accelerating existing inequalities. The trial doesn't blame technology. It points to the choice made by those who opted for this design knowing its effects. That choice now carries legal weight.

Big Tech's invincibility may be waning. Or this may just be the start of an era of prolonged appeals. Probably both. What can't be undone is that a jury validated that the problem was baked into the product's logic from the start. Not abusive users. Not neglectful parents. Intentional engineering of dependency. Once it's on the judicial record, that doesn't disappear. Legal precedents act like stones: they endure even when forgotten.

Stones don't lie, but historians sometimes do.

Sources:

1. LA NACION — New Mexico verdict against Meta, March 24, 2026

2. NPR — Internal Meta documents on strategy involving minors

3. RAPPLER — Analysis of the Section 230 breakthrough via focus on product design

4. Scimex — Mental illness social costs in the UK and Australian case law

5. FinancialContent — Judicial validation of the "addictive-by-design" theory and app architecture