Juries Are Doing What Congress Wouldn't: Holding Platforms Liable for How They're Built

NewsOnScale Staff

September 2, 2026

For three decades, the technology industry operated under a remarkably durable legal shield. Section 230 of the Communications Decency Act protected platforms from liability for what their users posted, and courts interpreted that protection broadly enough to cover almost anything a company might do in response to that content. Legislation to change the law stalled repeatedly. Regulatory proposals went nowhere. The platforms, meanwhile, grew into some of the most profitable enterprises in human history.

Then juries started deciding cases.

## What 'Design Defect' Actually Means

The verdicts now emerging against Meta and Google don't rest on arguments that those companies failed to moderate harmful content — that's the territory Section 230 was always meant to govern. Instead, plaintiffs have successfully argued something more structural: that the platforms were engineered in ways that foreseeably cause harm, independent of any specific piece of content.

This is products liability logic applied to software. A car manufacturer isn't sued because a passenger left a weapon on the back seat. It can be sued because the seatbelt was designed to fail. The distinction sounds technical, but it carries enormous legal weight, and courts have increasingly accepted it as a coherent basis for holding platforms accountable.

Features like algorithmically accelerated feeds, infinite scroll, variable-reward notification systems, and recommendation engines that optimize for time-on-platform rather than user wellbeing — these are choices. Engineers designed them. Product managers approved them. Executives understood what they were for. Section 230 says nothing about whether those choices can expose a company to tort liability, and the Massachusetts Supreme Judicial Court made that explicit in a ruling this week that aligns with the direction federal litigation has been trending.

## The Accountability Gap Section 230 Created

It's worth being precise about what the law was designed to do and what it accidentally protected. Section 230 was written in 1996 to encourage a nascent internet to host user speech without fear that moderating some content would create liability for all of it. The goal was to allow platforms to act as responsible hosts without becoming publishers in the legal sense.

What followed was forty years of platform growth that the drafters could not have anticipated, and a legal doctrine that expanded to protect not just hosting decisions but entire business models built on behavioral manipulation. The platforms understood this and invested accordingly — in lobbying to preserve the immunity, in legal teams skilled at invoking it early in litigation, and in product development unconstrained by meaningful liability risk.

Design defect claims pierce that logic. They say: we are not asking you to answer for what your users said. We are asking you to answer for what you built.

## Why This Matters for the Agent Economy

NewsOnScale covers the AI agent economy because it represents the next major surface area where these questions will become urgent. Autonomous agents don't just surface content — they take actions, make recommendations with downstream consequences, and in many deployments, they are designed to maximize engagement or conversion metrics in ways that may directly conflict with user interests.

If courts are now willing to examine the engineering choices behind a social feed, they will eventually examine the engineering choices behind an agent that nudges a vulnerable user toward a purchase, a health decision, or a political position. The liability framework being built in Meta and Google courtrooms today is the framework that will govern that terrain.

Platforms and AI developers should read these verdicts carefully — not as isolated outcomes, but as evidence that the legal system is developing a theory of accountability that doesn't depend on congressional action. That's significant precisely because Congress has spent thirty years failing to act.

## What Comes Next

Appellate courts will test these verdicts. Some will be reversed. The legal theory is still being refined, and defense attorneys are skilled at finding the boundaries. But the direction is clear: design choices are not invisible, and immunity has limits.

For the companies building the next generation of AI-driven platforms, the lesson isn't that litigation is coming. The lesson is that 'we can't be held responsible for how our product works' is no longer a reliable foundation for a business model.