The Algorithm Is an Editor. It Is Time Platforms Admitted It.

Nobody sits down and reads a newspaper anymore. Instead, the newspaper reads you. The average person now spends hours a day inside feeds curated by recommendation systems that decide, out of billions of posts, which few dozen deserve attention. These systems are not mirrors reflecting what we want. They are editors, making millions of publication decisions every second about what to promote, what to bury, and what to amplify. And unlike every editor in history, they answer to no one.

This is not a metaphor. In 2024, the United States Supreme Court ruled in Moody v. NetChoice that a platform's choices about what third-party content to display, and how to display it, are expressive editorial judgments protected by the First Amendment. The Court looked at Facebook's News Feed and YouTube's homepage and called them exactly what they are: editorial products. The algorithm is an editor, as a matter of law.

That should have settled something. Instead, it opened a contradiction that the tech industry has been exploiting ever since.

The editor who takes no responsibility

Consider what an editor does. An editor selects, arranges, promotes, and frames other people's work. That is also precisely what a recommendation algorithm does, except it does it at a scale no human editor could ever match, and it optimizes for a single metric: engagement. The algorithm promotes what keeps you scrolling. Outrage keeps you scrolling. Conspiracy theories keep you scrolling. A 14-year-old searching for diet tips gets recommended eating disorder content because the system has learned that despair is sticky.

Platforms have enthusiastically claimed the privileges of editorship. When governments tried to force them to carry content they did not want to carry, they invoked their editorial discretion, and the Supreme Court agreed. But when victims of algorithmic amplification ask for accountability, the same companies transform into neutral conduits and invoke Section 230, the 1996 law shielding them from liability for third-party content. They are editors when rights are at stake, and mere pipes when duties are.

They are editors when rights are at stake, and mere pipes when duties are. No other industry gets both.

No other industry gets both. A newspaper cannot claim First Amendment protection for its front page and then deny all responsibility for what it puts on the front page. Yet platforms do exactly that, every day, for the most-read front page in human history.

Courts are already fighting about it

The legal system is fracturing along exactly this fault line. In August 2024, the Third Circuit held in Anderson v. TikTok that TikTok's For You Page algorithm, which had recommended the lethal "Blackout Challenge" to a 10-year-old girl who died attempting it, was TikTok's own expressive activity, not third-party speech, and therefore outside Section 230's shield. If the recommendation is the platform's own speech, the court reasoned, the platform cannot disown it.

Meanwhile, the Fourth Circuit went the other way. In M.P. v. Meta, the court held that even algorithmic recommendations that allegedly radicalized a mass shooter were protected by Section 230's broad immunity, because Facebook had chosen to automate its editorial decision-making. The court openly conceded that the result was unsatisfying and declared that fixing it was "a question for Congress, not for judges." One court says the recommendation is the platform's own act. Another says the platform is immune for it. Both cannot be right, and the split is now begging for either Congress or the Supreme Court to intervene.

Regulate the amplification, not the speech

Here is where the usual debate goes wrong. The moment anyone proposes platform accountability, the industry warns of censorship, and in September 2026 the Trump administration's embassy in Canberra told Australia that its proposed algorithm opt-out rules could "facilitate censorship" by pressuring platforms to over-moderate. That objection has real force against laws that punish platforms for what users post. But it has no force at all against regulating what the platform itself chooses to amplify.

This is the distinction that matters: the post is the user's speech; the recommendation is the platform's. Nobody is asking to ban anything. The question is whether a company should be allowed to build a machine that hunts down vulnerable teenagers and feeds them content it knows will harm them, all in pursuit of watch time, and then claim it had nothing to do with it.

Australia's Digital Duty of Care bill, introduced to Parliament in October 2026, grasps this distinction. Its "My Feed, My Way" provision does not touch a single post. It simply requires platforms to offer users a genuine choice: an algorithmically recommended feed, or a chronological feed of accounts they actively follow, with penalties exceeding A$109 million for non-compliance. Europe's Digital Services Act has required very large platforms to offer a non-profiling feed option since August 2023. And this week, Meta rolled out new algorithm controls for Facebook globally, a concession that the demand for user control is not going away.

What editorial duty would actually look like

If we accept that the algorithm is an editor, the obligations follow naturally, and none of them require policing speech. First, users should be able to choose their editor: algorithmic or chronological, with the choice honored rather than buried three menus deep. Second, platforms should disclose the ranking signals that govern what gets amplified, the same way food labels disclose ingredients. Third, there should be a duty of care in design: when internal research shows a recommendation pattern causes foreseeable harm, the platform must fix the pattern, not just the occasional post it surfaces.

None of this tells anyone what they can say. It tells the most powerful publisher in history that publishing has consequences. The algorithm already exercises editorial power; the only question is whether it will ever carry editorial responsibility.

For three decades, the internet's bargain has been that platforms get the reach of a global publisher with the liability of a phone company. That bargain was written when a platform was a bulletin board. Today the platform is an editor with a billion readers, an editor that knows exactly what it is doing and can prove it with data. The law gave it an editor's rights in 2024. It is time the law asked for an editor's duties in return.