The Law That Built the Internet Turns 30 — and Washington Is Finally Asking Hard Questions

NewsOnScale Staff

July 11, 2026

In 1996, two congressmen inserted 26 words into a telecommunications bill that most of their colleagues never read carefully. Those words — "No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider" — became the legal scaffolding on which every major social media company, search engine, and user-generated content platform was eventually built.

This month, the Senate Commerce Committee held a hearing to mark the 30th anniversary of Section 230 of the Communications Decency Act. Judging by the transcript published by Tech Policy Press, the mood was not celebratory.

## What the Hearing Actually Revealed

The witnesses and senators who spoke over the course of the hearing reflected a rare cross-ideological consensus: that Section 230, as currently interpreted, insulates platforms from a level of accountability that its original authors almost certainly never intended.

The provision was drafted in response to a 1995 court ruling that punished a platform *more* for attempting to moderate content than for ignoring it entirely — a perverse incentive that Congress wanted to eliminate. The fix was sensible in context. What followed was decades of judicial interpretation that expanded the immunity far beyond content moderation decisions, effectively shielding platforms from liability for algorithmic amplification, design choices, and targeted recommendations.

That last category is now at the center of serious legal challenge. Courts in multiple jurisdictions — including, significantly, the Massachusetts Supreme Judicial Court in a ruling noted by privacy advocates this week — have begun drawing a distinction between a platform's role as a passive host of third-party content and its role as an active architect of user experience. Hosting a post is one thing. Engineering a recommendation engine that surfaces that post to a psychologically vulnerable teenager at 2 a.m. is something else.

## The AI Problem Nobody Has Solved

The hearing arrived at a moment when the Section 230 debate has collided with a second, more urgent policy emergency: the rapid deployment of AI-generated content at scale.

If the original law was designed to protect platforms from liability for what their *users* said, it was never designed to protect platforms from liability for what their own *systems* generate. AI content is not third-party speech. It is first-party output, produced by tools the platform built, trained, and deployed. Several legal scholars and advocacy organizations have now argued that Section 230 was never meant to — and should not — apply here.

Congress has been slow to act on this distinction, even as AI-generated disinformation, synthetic media, and automated manipulation campaigns have become standard features of the information environment. The 30-year anniversary hearing was, at minimum, a signal that some members of the Senate are aware the clock is running.

## What Accountability Would Actually Require

Reforming Section 230 without destroying the open internet is a genuine technical and legal challenge. Blanket removal of platform immunity would expose every comment section host to ruinous litigation and would almost certainly benefit large incumbents — who can afford legal departments — over smaller competitors and independent publishers.

But the current framework has its own market-distorting effects. When platforms bear no liability for the downstream consequences of their design choices, they have weakened incentives to invest in safety infrastructure. The public absorbs those externalized costs in the form of harassment campaigns, radicalization pipelines, and coordinated manipulation of civic discourse.

The most durable reform proposals tend to cluster around a middle path: preserving baseline immunity for passive hosting while removing it for algorithmic curation, design features proven to cause harm, and AI-generated content. That framework would require platforms to defend their product choices in court the way other product manufacturers do — not a radical concept, but a politically difficult one given the lobbying infrastructure the industry has built over 30 years.

## The Anniversary That Changed the Conversation

Hearings produce transcripts. Transcripts do not always produce legislation. But the Section 230 anniversary hearing matters because it reflects a shift in the baseline assumption. For most of the law's history, the default position in Washington was that platform immunity was a public good that required active defense. That default appears to be eroding.

What replaces it — careful surgical reform or something blunter and more politically motivated — remains the open question. The answer will shape not just who gets to speak online, but who bears responsibility when speech causes harm.