What's the Dead Internet Theory? Origins, Bots and AI Slop

What is the dead internet theory?

So what is the dead internet theory, in a sentence? It is the belief that most of the internet is no longer made by humans. Posts, comments, reviews, likes and even whole personalities are, according to the theory, largely generated by bots and automated systems, then pushed into feeds by algorithms that decide what you see.

That core idea comes in two layers, and keeping them apart is the key to understanding the whole debate:

• The observation: organic human activity online has been crowded out by bots, automated content and algorithmically curated results.
• The conspiracy: this crowding out is deliberate, a coordinated effort by governments, corporations or other powerful actors to manipulate the population and reduce genuine human contact.

The first layer is something researchers can measure, at least in part. The second is a claim with no documented evidence behind it. Much of the confusion around the theory comes from people using the same three words for both.

Two terms appear constantly in these discussions. A bot is software that performs automated tasks online, from harmless indexing to posting fake replies. AI slop is a slang term for low-effort, mass-produced content made with generative AI, the kind of images and text that tools like large language models (LLMs, the systems behind chatbots such as ChatGPT) can churn out in seconds.

Where the dead internet theory came from

Like many online folk beliefs, the theory’s birth is hard to pin down. Academic writing has trouble documenting fringe internet subcultures, and the earliest traces are scattered. The first posts on the idea are thought to have appeared on Wizardchan, a small imageboard, though that origin is not firmly documented.

The version most people know arrived in 2021. A user calling themselves “IlluminatiPirate” posted a long thread titled “Dead Internet Theory: Most Of The Internet Is Fake” on the esoteric board of a forum called Agora Road’s Macintosh Cafe. The poster said they were building on earlier threads from that board and from Wizardchan.

The post laid out its case in two parts. The “setup” was the claim that bots and automated content had displaced real people, and here the author leaned on a 2018 New York magazine article titled “How Much of the Internet Is Fake? Turns Out, a Lot of It, Actually.” The “thesis” went much further, stating that “the U.S. government is engaging in an artificial intelligence-powered gaslighting of the entire world population.”

From there the idea spread the way such ideas do: through YouTube explainers, Reddit threads and Twitter discussion. In September 2021, The Atlantic published “Maybe You Missed It, but the Internet ‘Died’ Five Years Ago,” which brought the theory to a mainstream audience and has been widely cited since.

The theory then crossed into scholarship. In 2023 a CRC Press book included a definition of it in its glossary. In 2024 an opinion piece called “Artificial influencers and the dead internet theory,” by Yoshija Walter, appeared in the journal AI & Society, focusing on AI-generated content rather than any plot. Few internet conspiracy theories make that journey, which says something about how much of it touched a real nerve. Readers who follow how ideas like the Hollow Earth theory moved from fringe pamphlets into wider culture will recognize the pattern, though this one did it in about three years.

Weak, strong and “leaner” versions of the theory

Because the idea grew in so many directions, writers have tried to sort its variants. Put the question “What is the dead internet theory?” to a forum regular, a journalist and a researcher, and you may get three different answers.

The weak and strong versions

A 2025 chapter in the book Market-Oriented Disinformation Research split the theory in two:

• The “weak” version: a group of elites is using bots to shape public discourse.
• The “strong” version: society itself has already collapsed after some catastrophic event, and an entity (possibly aliens or a highly advanced AI) keeps people plugged into the internet to hide that reality.

The strong version reads more like a science fiction premise than a claim about the world, and it sits comfortably alongside other reality-bending online ideas such as the Mandela Effect, where shared misremembering gets reframed as evidence that something about reality has shifted.

The “leaner” version

In a 2026 article in the publication Computer, Hal Berghel described a “leaner” dead internet theory, one stripped of the paranoia and politics. Building on Walter’s 2024 piece, he identified the core as four things: algorithmically generated content, the byproducts of generative AI, the difficulty many people have telling those apart from human work, and the mistrust and misinformation that follow. His argument was that conspiracy theorists take real phenomena and stretch them into implausible claims, but that the underlying criticisms should not be dismissed.

This leaner reading is closest to how the phrase gets used today in casual conversation. When someone on Reddit replies “dead internet theory” under a suspiciously generic thread, they rarely mean the government is gaslighting the planet. They mean: these look like bots.

The evidence people point to

Supporters and commentators lean on a mix of hard numbers and vivid examples. Some of it is solid. Some of it is anecdote that feels like data. The honest reply to “What is the dead internet theory based on?” is: both.

How much web traffic is bots?

The most cited figure comes from the security firm Imperva. In 2016 it examined more than 16.7 billion visits to 100,000 randomly chosen domains and found that automated programs accounted for 52% of web traffic. Its report covering 2023 put automated traffic at 49.6%, up about two percentage points from the previous year, an increase partly attributed to AI companies scraping the web for training material. By 2026, Fortune reported that more than half of all website requests came from bots.

These numbers are real, but they measure something specific: requests hitting servers. A search engine crawler indexing a page counts. So does a price-monitoring script or a security scanner. Traffic is not the same as conversation, and the figures do not show that half of all posts or replies are written by machines.

Chatbots and the AI boom

The public release of ChatGPT in late 2022 changed the tone of the conversation. Before it, the theory mostly imagined governments, corporations or skilled insiders running the bots. Afterward, anyone could produce fluent text at scale, and journalists began describing the theory as more realistic than before.

Predictions added fuel. In 2022 Timothy Shoup of the Copenhagen Institute for Futures Studies said that if GPT-3 “gets loose,” the internet would be “completely unrecognizable,” and suggested 99% to 99.9% of online content could be AI-generated by 2025 to 2030. That was a forecast, not a measurement, yet it is often repeated as if it were evidence.

In September 2025, OpenAI chief executive Sam Altman posted on X that he “never took the dead internet theory that seriously but it seems like there are really a lot of LLM-run twitter accounts now.” The post went viral, partly because of who wrote it.

Social media and AI slop

Several episodes became shorthand for the feeling:

• “I hate texting” (2020): a wave of Twitter accounts posted near-identical tweets such as “i hate texting i just want to hold ur hand,” collecting tens of thousands of likes, many suspected to come from bots.
• Shrimp Jesus (2024): AI-generated images flooded Facebook, including depictions of Christ fused with shrimp, many drawing hundreds or thousands of comments reading “Amen.” Writer Thomas Sommerer treated Shrimp Jesus as a symbol of the whole shift.
• Meta’s AI accounts (January 2025): Meta executive Connor Hayes described AI personas that would “exist on our platforms, kind of in the same way that accounts do,” with bios and profile pictures. The announcement renewed interest in the theory almost overnight.
• SocialAI (2024): an app launched by Michael Sayman in September 2024 let users post to an audience made entirely of AI bots, a deliberate, opt-in version of the dead internet.
• Digg (2026): Reddit co-founder Alexis Ohanian said in 2025 that he had long subscribed to the theory. He and Digg founder Kevin Rose relaunched Digg in open beta on January 14, 2026; it closed on March 14, citing an “unprecedented bot problem” among other issues, and was rebooted in May as an AI-driven news aggregator.

Search engines, link rot and YouTube’s “Inversion”

Proponents also argue that search engines show far less of the web than they claim, filtering out content and leaving a curated shell. Link rot, the steady decay of old links as pages vanish, makes the web feel smaller and emptier over time. In 2024 Google itself acknowledged that its results were being flooded with sites that “feel like they were created for search engines instead of people,” and a spokesperson recognized generative AI’s role in that flood.

On YouTube, fake views generated by software were once so common that some engineers worried the detection system would start treating fake views as normal and flag real ones as suspicious. They called this hypothetical tipping point “the Inversion.” It is one of the most quoted details in dead internet lore, because it captures the fear precisely: a system that can no longer tell humans from machines.

What is documented and what is claimed

Laying the pieces side by side gives a practical answer to the central question: what is the dead internet theory, really? Part measurable trend, part myth.

The left column is uncomfortable enough on its own. The right column is where the conspiracy lives, and none of it has been shown.

What skeptics and researchers say

The expert view is fairly consistent. Bot accounts and automated traffic make up a large share of internet activity, but that does not prove a coordinated plan to replace human beings online.

Critics point out that much of the theory rests on impressions rather than data: a sense that content has gotten worse, that replies feel repetitive, that the old web was livelier. Those feelings may be accurate descriptions of experience, but they are not measurements. Nostalgia may play a part too, since memory tends to polish the past.

Researchers also note that bot activity has many mundane purposes. Search crawlers, spam filters and data scrapers all count as automated traffic. Some bots are malicious, such as those used in denial-of-service attacks (floods of requests meant to knock a site offline), but even those are about disruption or profit, not replacing people.

Scholars who compare it with other internet conspiracy theories see a familiar move: take a real trend, like AI or rising bot activity, and extend it into a sweeping claim without concrete proof. For more on how that pattern plays out elsewhere, the journal’s conspiracy theories section looks at several cases where a kernel of truth grew into something much larger.

There is a real risk the leaner critics flag, though. Toby Walsh, a computer scientist at the University of New South Wales, has warned that training new AI systems on content made by older AI could degrade quality over time. Others have described bots replying to bots in loops that only a human could break. That is not a conspiracy, just an ecosystem eating its own output.

Why the dead internet theory persists

Few modern conspiracy theories have been partially vindicated by events the way this one has. The 2021 post sounded paranoid; by 2025, the head of the company behind ChatGPT was half-joking about it in public. That arc gives the idea a strange credibility, even though the core conspiracy remains unsupported.

It also speaks to something older than computers. Stories about a world that looks normal but is quietly hollow, filled with doubles or replacements, run deep in folklore and fiction. The dead internet is a digital cousin of the changeling tale: the familiar face that is not quite right. That uncanny feeling belongs to the territory the journal explores under consciousness and the mind, where perception and reality do not always line up.

Finally, the theory gives people language for a real frustration. Asking “What is the dead internet theory?” is often a way of asking a different question: why does the internet feel lonelier than it used to? The theory supplies a villain. The evidence suggests something less cinematic, a web shaped by advertising incentives, cheap automation and algorithms that reward whatever keeps people scrolling.

How to spot the dead internet in your own feed

Whatever you make of the conspiracy, the theory carries a practical lesson: learn to notice automated content. A few habits help.

1. Check the account, not just the post. Brand-new profiles, generic names with number strings and feeds full of identical-style replies are common bot signs.
2. Watch for template replies. Comments that praise without engaging (“Amazing! So true!”) or restate the post’s own words often come from automation.
3. Look closely at images. AI images frequently have warped hands, melted text, impossible details or the polished, airless look of Shrimp Jesus.
4. Notice engagement that does not match. A bland post with enormous likes and few real conversations deserves suspicion.
5. Favor smaller, moderated spaces. Communities with active human moderators tend to hold up better than open feeds.

The common mistake runs in both directions. Some people dismiss every complaint about bots as paranoia, ignoring documented trends. Others start treating every disagreeable stranger as a bot, which is its own kind of distortion, and an easy way to stop listening to real people. The record supports caution, not certainty.

So, one last time, what is the dead internet theory? It is a 2021 forum conspiracy that guessed wrong about the plotters and arguably right about the direction of travel. The machines are not running a coordinated hoax, but they are filling more of the web every year, and the question of how to find each other in that noise is still open. Readers who enjoy chewing on questions like this can browse the full journal, or settle into the Troubled Minds Vault for long-form conversations about the unexplained, made by humans and still very much alive.

Questions and answers

What is the dead internet theory in simple terms?

It is the idea that much of what you see online, from comments and likes to posts and whole accounts, is produced by bots and AI rather than real people. The original version adds a conspiracy: that governments or corporations did this on purpose to steer public opinion. The first part has real data behind it; the second part has no supporting evidence.

Is the internet 51% bots?

Roughly, if you mean traffic. Security firm Imperva found automated programs made up 52% of web traffic in its 2016 analysis and 49.6% in its report covering 2023, and by 2026 more than half of website requests were reported to come from bots. That measures requests to servers, though, not the share of posts or conversations written by machines, which is much harder to count.

Why are people saying the internet is dying?

Many people feel online spaces have become crowded with spam, engagement bait, AI-generated images and accounts that post like machines. Viral oddities such as the Shrimp Jesus images on Facebook, Meta's 2025 plan for AI-run accounts, and a relaunched Digg shutting down over bot problems in 2026 all fed the sense that real human conversation is being drowned out.

What does dead internet theory mean on Reddit and in slang?

In everyday use, especially on Reddit and social media, calling something 'dead internet theory' is shorthand for 'this looks like bots talking to bots.' People drop it under suspicious threads, repetitive replies or AI-generated posts. The slang usage usually means only the observation that automated content is everywhere, without the full government conspiracy attached.

Who started the dead internet theory?

The idea is thought to have first appeared on the imageboard Wizardchan. It spread widely in 2021 when a user called IlluminatiPirate posted 'Dead Internet Theory: Most Of The Internet Is Fake' on the forum Agora Road's Macintosh Cafe. A September 2021 article in The Atlantic then brought it to a mainstream audience.

Will the internet exist in 50 years?

Nothing in the record suggests the internet itself will disappear. The open question is what it will feel like: how much content will be machine-made, and whether platforms can verify real people. Some predictions about AI flooding the web have been dramatic, but long-range forecasts are speculation, and the infrastructure of the internet is not what the theory says is dying.

Stay curious. Keep the question open.
Follow the evidence wherever it leads.