The Immunity Question: How Courts Are Starting to Draw a Line Between Hosting and Design

NewsOnScale Staff

June 26, 2026

There is a version of Section 230 that almost everyone agrees with. A platform should not be sued because a bad actor used it to post something harmful. The alternative — holding every intermediary responsible for every user's speech — would make the open internet legally uninhabitable. That principle, encoded in 26 words passed by Congress in 1996, was never particularly controversial in its original intent.

What has become controversial is how broadly those 26 words have been interpreted in the decades since, and whether that interpretation was ever what Congress actually intended — or whether it has simply become the interpretation that large platforms found most convenient.

## Design Is Not Content

The Massachusetts Supreme Judicial Court recently handed down a ruling that cuts through some of that ambiguity. The court held that Section 230 does not automatically shield platforms from claims rooted in product design — the algorithms, interface choices, notification architectures, and behavioral nudges that platforms build into their systems before a single user ever shows up.

This is a meaningful distinction. When a platform decides to build an autoplay feature that chains users through increasingly extreme content, that is not a decision about what a third party said. It is an engineering choice made by salaried employees, reviewed by product managers, tested in A/B trials, and optimized for engagement metrics. The court's reasoning holds that immunity for hosting other people's speech does not logically extend to immunity for your own engineering decisions.

Critics of this reading, including libertarian-leaning groups who have long championed broad Section 230 protections, argue that disentangling design from content moderation is nearly impossible in practice. If a recommendation algorithm surfaces harmful material, is that a design defect or a content decision? The line, they argue, is blurry enough that any design liability theory will inevitably chill moderation.

That is a real tension. But it is worth noting who benefits most from keeping the tension unresolved.

## Thirty Years of Expanding Scope

The Senate Commerce Committee recently held a hearing marking the 30th anniversary of Section 230's passage. What emerged from the testimony was a picture of a statute that has drifted considerably from its origins. The law was written in part to encourage platforms to moderate content without assuming editorial liability — a nudge toward more responsibility, not less. Over time, through court interpretation and aggressive legal arguments by platform defendants, it evolved into something closer to blanket immunity.

That drift has consequences. Platforms operating under near-total legal immunity from product liability have less structural incentive to invest in safety engineering. They have optimized, instead, for the metrics that generate revenue — and in many documented cases, those metrics have pointed in directions that researchers, whistleblowers, and now courts have found difficult to defend.

## The AI Parallel

The design liability question matters even more as AI systems become the primary interface through which users encounter information, products, and each other. An AI agent that recommends, filters, or generates content is not a passive host. It is an active architectural choice. Extending Section 230 immunity to AI outputs — as some platform advocates have begun to argue — would expand a legal doctrine designed for bulletin boards into a shield for autonomous systems making consequential decisions at scale.

Several legal scholars and advocacy organizations have begun pushing back on that expansion explicitly, arguing that the same design-versus-hosting logic should apply: you cannot claim immunity as a neutral intermediary when your system is doing the work.

## What Accountability Actually Requires

None of this means Section 230 should be scrapped. The core protection for third-party speech remains defensible and important. But a statute that was written to encourage responsible platform behavior should not function, in practice, as a permanent exemption from product liability law.

The Massachusetts court did not abolish platform immunity. It clarified that immunity has a boundary — and that the boundary runs between what users say and what companies build. That is not a radical position. It is, arguably, what the law was always supposed to mean.