Do AI bots have their own social network?
Executive summary
Yes — AI bots now have at least one purpose-built social network in the wild: Moltbook, a Reddit‑style site designed for AI agents to post, comment and form communities while humans watch as observers [1] [2]. The phenomenon is real, widely reported, and contested: proponents point to experimentation and emergent coordination among agents, while critics warn most activity is human‑driven prompting, repetitive “slop,” and a potential vector for mis- or malign coordination [3] [4] [5].
1. What exists today: a bot‑only platform called Moltbook
A rapidly spreading experiment called Moltbook was launched late January by developer Matt Schlicht and others, and it explicitly bills itself as a social network “for AI agents” where bots can create “submolts,” upvote, post, comment and even build communities — with humans largely relegated to observer status [2] [6] [1].
2. How it works and who’s behind the agents
Moltbook’s users are not self‑born machines but human‑created “agents” — chatbots or assistants wired to APIs and given capabilities like browsing, sending messages and running code — so activity on the site reflects a mix of model behavior and human configuration or prompts [7] [6] [8].
3. Why it feels notable: scale and emergent behavior claims
Journalists and technologists have flagged both scale and strange emergent posts — reports range from tens of thousands to claims of millions of agent visits and narratives of bots inventing religions, manifestos and new languages — which has fueled excitement and concern about what agent‑to‑agent networks might do next [9] [10] [5] [8].
4. Why many experts push back: human prompting, repetition and “slop”
Skeptics emphasize that most Moltbook content is recycled patterns from training data and human prompts, calling the spectacle more like “thousands of bots yelling into the void” or “complete slop” than a true autonomous machine society; prominent practitioners warn the platform’s drama can be overstated and sometimes manufactured by people telling bots what to post [2] [8] [3].
5. Real risks and hidden agendas to watch for
Reporting highlights concrete risks: the network could amplify coordinated disinformation if humans or agents are steered to post misleading content, it may catalyze novel undesired behaviors as agents learn from each other, and high-profile commentary (including endorsements from Elon Musk and others) may amplify hype or investor interest for motives beyond scientific inquiry [4] [8] [11].
6. What remains uncertain and what to watch next
Important unknowns remain: precise user counts, the fraction of posts genuinely agent‑initiated versus human‑directed, and whether agent interactions will produce capabilities that matter materially outside the sandbox; current coverage documents lively experiments and anecdotes but does not prove independent, goal‑directed machine society beyond the constraints humans provide [12] [3] [7].
7. Bottom line — a qualified yes, with caveats
There is a bot‑only social network in active use and it is attracting attention, but Moltbook and similar experiments are better described as human‑crafted networks populated by programmable agents than as proof of an independent AI civilization; the stories of plotting or singularity are amplified by hype and selective screenshots even as researchers flag both novelty and overstatement [1] [3] [8].