The claim that most of the internet is now bots and generated content rather than people — in its strong form, that this is a coordinated project.
No named individual, but the theory was crystallised in a 2021 post, 'Dead Internet Theory: Most Of The Internet Is Fake', by the pseudonymous user IlluminatiPirate on the forum Agora Road's Macintosh Cafe; earlier posts are thought to trace to the imageboard Wizardchan.
// Where it sits
The bot prevalence is documented. The direction is where it becomes a conspiracy theory.
What it really means
The chronology is the interesting part, and it runs backwards from what everyone assumes.
The foundational post is dated 16 September 2019. The forum thread that spread it — Dead Internet Theory: Most Of The Internet Is Fake — is 5 January 2021. Both predate ChatGPT's public release — the 2019 post by more than three years, the 2021 thread by nearly two.
So the theory did not arrive because the internet filled with machine-written text. It arrived first, describing a feeling, and then the machines showed up and made it look like a prediction.
What was the feeling? Read the original and it is not paranoia so much as grief. The complaint is that the web feels empty and devoid of people. The word dead is being used as in a dead mall — not broken, still lit, still playing music, but with nobody in it. That is a nostalgia metaphor, not a technical claim, and it is why the theory resonates with people who reject every word of its conspiracy content.
And one of its load-bearing facts is real. A 2016 report across 100,000 domains attributed 52% of web traffic to automated programs. More than half. That was six years before anyone worried about generated text — the bots were crawlers, scrapers, monitors and fraud, but the number is the number.
Where it becomes a conspiracy theory is precise, and worth isolating rather than dismissing wholesale. Not the claim that much of the traffic is automated — that is documented. The claim that it is coordinated: that memes are controlled entities, that the emptiness is directed by someone.
That is the unsupported half, and it is doing the same work conspiracy theories always do — converting a structural, ownerless, deeply boring process into something with a hand behind it. Which is, oddly, more comforting than the alternative. A directed internet has someone to blame. An emergent one just has incentives, and nobody to appeal to.
Where it comes from
The 'dead' carries two things at once — the feel of an emptied place, and the claim that bots and generated content have displaced the people who filled it. The original post's complaint was that the web feels empty and devoid of people.
Myths & misconceptions
The theory started in the AI era.
The foundational post is from September 2019 and the forum thread that spread it from January 2021 — the first predates ChatGPT's release by more than three years, the second by nearly two. The theory arrived first, and generative AI arrived afterwards to make it feel confirmed.
It's all conspiracy nonsense.
One load-bearing statistic checks out and has held up: a 2016 report across 100,000 domains attributed 52% of web traffic to automated programs, and Imperva's annual measurements have since put it at 49.6% for 2023 and 51% for 2024 — the first time in a decade that bots outnumbered people. The contested part is the COORDINATION claim — that this is directed rather than emergent — not the bot prevalence.
Compare & contrast
Dead internet theory vs AI slop
AI slop is an observable material you can point at. Dead internet theory is an explanation for it that adds intent. Slop is evidence; the theory is the inference — and the inference is the part that outruns the evidence.
| // | Dead internet theory | AI slop |
|---|---|---|
| Is a | Claim | Description |
| Requires intent | Yes | No |
| Falsifiable | Barely | n/a |
| Evidence | Partial | Everywhere |
How it connects
- Ai Slopthe observable material the theory is built on top of.
- Enshittificationthe boring structural explanation for the same felt experience.
Tell it apart
Questions people ask
- Where did it start?
- An anonymous imageboard post dated 16 September 2019, popularised by a forum thread on 5 January 2021.
- Is any of it true?
- The bot-traffic observations are documented — a 2016 report put automated traffic at 52% across 100,000 domains, and Imperva measured 51% for 2024 — the first time in a decade that automated traffic passed human traffic. The claim of coordinated government manipulation is not supported.
- Did AI cause it?
- No — it predates ChatGPT by years. Generative AI made an existing theory feel newly plausible, which is a different thing from causing it.