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

NewsOnScale Staff

September 16, 2026

There is a version of the platform accountability debate that has played out in congressional hearings for the better part of a decade: senators display screenshots, executives apologize with calibrated contrition, and nothing binding happens. That version appears to be losing relevance. The courts are moving.

Recent verdicts against Meta and Google — targeting not what users posted, but how platform systems were engineered to maximize engagement — represent a meaningful shift in how American law is beginning to treat the relationship between product design and public harm. These are not defamation cases or misinformation disputes. They are product liability cases, and that distinction matters enormously.

## What Section 230 Actually Protects

Section 230 of the Communications Decency Act grants online platforms broad immunity from liability for third-party content. The logic, written in 1996, was straightforward: you shouldn't sue a telephone company because someone said something harmful over their lines. Platforms aren't the speakers, so they shouldn't bear the speaker's legal exposure.

What Section 230 does not protect — and what courts are now scrutinizing — is the platform itself as a designed object. An algorithm that learns a teenager is emotionally vulnerable and then serves escalating content to exploit that vulnerability is not a passive conduit. It is a product making active choices. The legal argument gaining traction is that those choices belong to the company's engineers and executives, not to any individual user who uploaded a video or wrote a post.

This framing has been developing in academic and advocacy circles for years. Its arrival in verdict form is something different.

## The Design Defense Collapses

Meta and Google have argued, in various forms, that holding them liable for algorithmic recommendations would effectively punish them for hosting speech — that you cannot separate the recommendation engine from the content it recommends. Courts delivering verdicts against this position are saying: yes, actually, you can.

The practical implications are significant. If a platform's engagement-optimization system is treated as a product subject to standard design-defect analysis, then companies face the same questions any manufacturer faces: Did you know this feature caused harm? Did you have safer alternatives? Did you ship it anyway?

Internal documents that have surfaced in various proceedings suggest that in several cases, the answer to all three questions may be yes. That is not a Section 230 question. That is a corporate accountability question, and American tort law has tools for it.

## What This Means for the Agent Economy

The timing is pointed. As AI agents increasingly mediate what users see, buy, read, and decide — operating with far more autonomy than a static recommendation algorithm — the question of where platform immunity ends becomes more urgent, not less.

An AI agent that steers a user toward a financial product, a medical decision, or a political position is not simply hosting content someone else created. It is acting. The legal frameworks being tested against Meta and Google today will almost certainly be applied to AI-driven platforms tomorrow, and the companies building those systems would be unwise to assume otherwise.

The Institute for Family Studies and Public Knowledge, coming from very different ideological orientations, have both recently argued that AI should not inherit the Section 230 protections that social platforms have relied upon. When organizations that rarely agree on anything reach the same structural conclusion, it is worth paying attention.

## The Gap Congress Left Open

Legislative reform of Section 230 has stalled repeatedly, caught between conservatives who want to punish platforms for alleged anti-conservative bias and progressives who want to punish them for amplifying harmful content. The two critiques require nearly opposite legal remedies, and the impasse has been durable.

Litigation does not wait for consensus. The verdicts accumulating against major platforms are filling a regulatory vacuum that Washington has repeatedly declined to address. Whether that is the right mechanism for technology governance is a legitimate debate. That it is happening — and accelerating — is no longer a prediction. It is the current state of play.