By Saihaj Sraon
Saihaj is a member of the ConnectSafely Youth Advisory Council.
In 1996, Congress amended the original Communications Act of 1934 with the Communications Decency Act. Embedded in that amendment is Section 230, widely hailed as one of the most consequential pieces of internet legislation in U.S. history. At its core, Section 230 reads:
“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.”
This provision gave birth to the modern social-media ecosystem. Imagine a world where anyone and everyone could sue Meta, or their online followers, simply because they disliked a post. The prospect of billions of users, corporations, or even hostile governments, filing lawsuits over online speech would have stifled innovation and choked free expression. By insulating platforms and users from liability for third-party content, Section 230 became a bastion for the fundamental right to free speech on the World Wide Web. But the web has since evolved.
From Town Squares to Algorithmic Megaphones
Early message boards were literal town squares: anyone could say something, anyone could listen, and every contribution received equal treatment. That egalitarian digital plaza is exactly the environment Section 230 was designed to protect.
When platforms introduced recommendation algorithms, the experience changed dramatically. On nearly all popular social media sites, the algorithm acts like a megaphone, amplifying some voices while drowning out others. It does this based not on random chance but on a user’s unique digital footprint. The result is a town square where every user hears something different. The once-open space has now become an echo chamber where selected voices roar and others fade into silence depending upon who you are.
These megaphones are also engineered to keep users glued to the platform, often at the expense of mental health. By harvesting massive troves of data, companies can profile users based on age, income, spending habits, health status, mental health conditions, and countless other factors. They quickly learned that the most effective way to sustain attention is to serve each profiled user a steady stream of cherry-picked posts, content that resonates, even when it is addictive or detrimental to them specifically.
The relentless, algorithmically curated flood of content is especially damaging to children and teens, whose brains are still in critical stages of development. Continuous exposure to such curated material has been linked to higher rates of anxiety, depression, and, in severe cases, acute mental-health crises. In addition to lasting psychological challenges in young people, adults can also experience a gradual erosion of life satisfaction. Already vulnerable populations can be pushed toward self-harm.
Section 230 grants platforms immunity for the algorithmic recommendation systems that repeatedly surface certain posts in targeted streams. The statute’s original design assumes that harmful content originates solely from users, overlooking the fact that the algorithm, rather than any individual post, is often the true engine of the distress described above.
An Amendment to Section 230
When an algorithm negligently surfaces material that inflicts tangible harm, the platform should be accountable for that injury, while the original poster remains protected because they have the right to free expression and because they lack control over the algorithm’s amplification. A clear line must be drawn between speech (a user’s contribution to the platform) and algorithmic technology (the platform’s selection and targeted amplification of certain speech). The proposed amendment to Section 230 should:
- Retain Section 230 immunity for user-generated content, preserving the free-speech protections that made online expression possible.
- Remove algorithmic recommendation systems from that blanket immunity. If an algorithm is negligently designed or operated and causes demonstrable harm, the platform can be held liable by the injured party.
This approach would compel companies to embed safeguards into their algorithms, protecting users while still allowing the free exchange of ideas.
Section 230 was a visionary statute that helped transform the internet into a democratic public square. Yet that square has been reshaped by powerful algorithms that decide which voices are amplified, which are silenced, and who gets to hear them. Because these algorithms can inflict genuine psychological injury, especially on children, teens, and other vulnerable groups, extending blanket immunity to them is no longer defensible.
An amendment that preserves Section 230’s shield for user-generated speech while stripping that protection from platform algorithms whenever they cause verifiable injuries strikes the right balance. The internet would remain a place for free expression, and companies will be answerable for the engineered harms propagated by their technology. By turning down the volume on the megaphones and targeting the code that strategically amplifies content, rather than the content itself, we can protect both liberty and well-being online. In short, regulate the algorithm, not the post.
The opinions expressed in this post are those of the author and some other members of ConnectSafely’s Youth Advisory Council. They do not necessarily represent the views of the Council as a whole or ConnectSafely.