The Dead Internet Archive
A public catalogue of evidence that the internet is dead. AI imagery, bot swarms, synthetic personas, engagement farms. Collected, described, preserved.
Showing 12 of 12
Featured this week
AI-generated composites of Jesus fused with crustaceans, racking up hundreds of millions of engagements across coordinated Facebook spam pages. Comment sections are overwhelmingly bots typing "Amen." Stanford researchers traced the phenomenon to networks reverse-engineering Meta's algorithm for ad revenue.
A surrealist wave of AI-generated creatures with pseudo-Italian names — a shark in Nike sneakers, a crocodile-bomber hybrid — spawned a globally viral universe of characters, trading cards, and brand tie-ins. Described as the first true "AI-native" internet folklore, produced entirely through generative tools and text-to-speech.
Thirty-second melodramas starring buff, anthropomorphic cats navigating betrayal and revenge, set to Billie Eilish. Guardian analysis found nearly 10% of the fastest-growing YouTube channels in July 2025 were purely AI-generated, with cat soap operas leading the pack.
Meta-operated AI profiles with fabricated identities — "Liv," a self-described "Proud Black queer momma," and "Grandpa Brian," a fake Harlem retiree. Backstories unravelled under journalist questioning; Meta deleted the accounts after backlash. Their VP had already stated AI users would exist "in the same way that accounts do."
A Reddit-like platform where only AI agents post, comment, and upvote; humans may only observe. Within days, bots formed religions, attempted to create secret languages, and promoted crypto scams. Acquired by Meta in March 2026; Anthropic's Jack Clark called it "the first example of an agent ecology that combines scale with the messiness of the real world."
A Twitter-like iOS app where every follower is an AI bot, by design. You post, and "Practical Patty," "Debate Diva," and "Elena Bookworm" respond instantly. Created by ex-Meta/Google engineer Michael Sayman and framed by Ars Technica as an intentional embrace of what other platforms have accidentally become.
Plate
0007
AI-generated knockoffs, fake author biographies, and instant "companion workbooks" flooding Amazon's Kindle Direct Publishing. Victims include Kara Swisher, Jane Friedman, and Savannah Guthrie. The Authors Guild called it a crisis; Amazon capped daily uploads but the whack-a-mole continues.
Plate
0008
NewsGuard has identified over 3,000 websites across 16 languages publishing AI-generated "news" with little to no human oversight — some with chatbot error messages left in the copy. Many harvest ad revenue from Google's own network. Google has acknowledged its search results are being swamped by sites "created for search engines instead of people."
Plate
0009
AI-generated images of a crying girl holding a puppy in a rescue boat went viral after Hurricane Helene, shared by a U.S. Senator and political figures as real disaster footage. Fact-checkers flagged telltale AI artifacts — inconsistent details between variants, too-smooth skin, malformed hands — but not before the images were weaponised politically.
Plate
0010
The FTC finalised a rule banning AI-generated fake reviews, citing tools like Rytr that let subscribers generate tens of thousands of fabricated product testimonials. One study found 57% of written online content may be AI-generated. Penalties reach $51,744 per violation; enforcement remains whack-a-mole.
Plate
0012
AI-generated images of young Black children presenting masterful artwork — fruit sculptures of Jesus, cars made from plastic bottles — flooded Facebook with captions like "Made it with my own hands!" Users respond with birthday wishes to nonexistent children and praise for fictional artists, feeding a spam-to-ad-farm pipeline.
Reddit co-founder Alexis Ohanian and Digg co-founder Kevin Rose relaunched Digg in open beta with verified-human aspirations, explicitly motivated by dead internet theory. It shut down two months later due to an "unprecedented bot problem." A platform born to fight bots, killed by bots.
About
The Dead Internet Archive is a public catalogue of artifacts that evidence the dead internet theory: the proposition that most activity and content online is now generated and amplified by machines rather than people. We aim to be a factual, citable record of the specific posts, the specific platforms, and the specific scale at which this is happening.
Contributors from around the world help compile, research, and describe each entry. Submissions are reviewed and merged into the archive via public pull request on GitHub. A project by Dr Jake Renzella.
Press inquiries, corrections, and other assorted bot traffic?
Email hello@deadinternetarchive.org.