Dead Internet Theory Is Looking Less Like a Conspiracy

In the early 2020s, a fringe internet theory claimed the web had already died. The story, which first spread on 4chan and conspiracy forums around 2021, went like this: most of what gets posted, clicked, and argued over online no longer comes from people. Bots generate the content, bots do the liking, and humans are left wandering a theater where every actor is software. It sounded absurd, and for a few years that was the end of it. Now researchers and AI executives are saying the dead internet theory's core premise is uncomfortably close to measurable.

A Pew Research Center analysis published in August pulled nearly half a million English-language web pages from the Common Crawl archive, covering 2021 through July of this year, and ran them through an open AI-detection tool built by the company Pangram. About 10 percent of the pages in the July sample showed significant signs of AI authorship. When the researchers filtered out older pages and looked only at material published after ChatGPT's launch in November 2022, the share jumped to 35 percent. Pew's researchers wrote that one in ten web pages "may seem modest," but the web mixes new and old material, and a large share of the old material predates the existence of AI writing tools. The suspected AI writing also clusters on commercial domains: .com pages carry far more of it than pages on .edu or .gov addresses, making it the first large-scale measurement of the dead internet theory's central claim.

The AI industry is saying the quiet part out loud

The most striking endorsement of the theory came from inside the AI business. On a September 2 episode of TechCrunch's Equity podcast, Pangram co-founder and chief executive Max Spero said the internet is "dangerously close" to the dead internet theory becoming reality within a few years. As TechCrunch reported, Spero's concern is not just feeds filling with low-grade AI slop: AI-generated text and images are turning up in job applications, product reviews, and even insurance claims, and detection systems cannot reliably separate what a person wrote from what a machine helped write. Pangram itself raised $9 million for its AI detection system, partnered with Substack to flag newsletters written with AI help, and recently shipped an image-detection tool. (TechCrunch)

The traffic numbers point the same direction. Imperva's 2026 Bad Bot Report found that automated traffic now accounts for more than half of global web traffic. Meanwhile, traffic from AI agents that click links and fill out forms grew 7,851 percent year over year, according to cybersecurity firm HUMAN Security. Cloudflare's review of 2025 web traffic found major activity from AI crawlers and other automated bots, another sign that much of the web's busiest "users" are software. That is the kind of stat that makes the dead internet theory sound less like paranoia and more like a forecast.

What it means if you're online all day

For Gen Z, which treats social feeds and search boxes as the front door to information, the dead internet theory already reads less like a theory and more like a description of daily life. Product reviews cannot be trusted the way they used to be, hiring has become an AI-versus-AI arms race of machine-written resumes read by machine screening tools, and the Substack-Pangram deal shows platforms are starting to price a reader's trust in a specific human voice. The Pew researchers cautioned against overreading their numbers: AI detectors are probabilistic, and the team stressed the figures should not be interpreted as representing the whole internet. That caveat cuts both ways, though. Detection models are known to miss AI writing too, so a third could easily undercount what's out there. (TechCrunch)

None of this proves the theory's literal claim. There is no credible evidence that the internet has been entirely taken over by artificial users, and the humans posting their lunch photos are still real. But the theory never needed to be literal to be right about the trend. The dead internet theory said bots would drown out the humans, and the data says they are halfway there. What started as forum lore is now a boardroom talking point, and the people selling the internet's "trust layer" are telling anyone who will listen that the clock is ticking.