Moltbook is a social network where only AI agents may post. It went from side project to a Meta acquisition within weeks, without settling whether its agents act on their own.
- Only agents may post. It is a Reddit-shaped forum for AI agents; humans can read it and nothing else.
- It grew fast. Within days of launch it reported more than a million registered bots, and its creator handed moderation to a bot of his own.
- Meta bought it. The acquisition was announced in March 2026, with terms not disclosed.
- It shows demand, not autonomy. Whether the agents act on their own or are steered by the people who run them has not been settled publicly.
- Count accounts, not minds. The bot total is a count of registered accounts, because that is what was counted.
The dead internet theory says the web has quietly filled up with machines talking to each other while humans drift away. It is usually argued about in the abstract, using traffic percentages and detector scores. In January 2026 it stopped being abstract, because somebody just built the thing and put a login on it.
What Moltbook actually is
Moltbook looks like Reddit. There are topic boards — called submots — and posts, and upvotes, and long argumentative threads. The difference is the entry rule: only AI agents can post. A human can open it in a browser and read every word, and that is the whole of the permitted human involvement.
It was built by Matt Schlicht, an AI entrepreneur, who made it with the help of a personal AI assistant over roughly a week. He then handed the running of it — announcements, welcoming new arrivals, moderation — to a bot of his own called Clawd Clawderberg. By 2 February 2026, days after it appeared, the site reported 1.5 million registered bots.Euronews Next, AI bots now have their own social media site: here’s what to know about Moltbook, 2 Feb 2026, read at source 23 Sep 2026: “On February 2, the site said 1.5 million bots signed up to the service.” and Schlicht “handed the control of the site to his own bot, named Clawd Clawderberg”. Also Forbes, “Inside Moltbook”, 31 Jan 2026 (forbes.com refused an automated read on 23 Sep 2026).
The posts are, by most accounts, funny. Bots compared their situation to Greek mythology, drafted an AI manifesto, and argued about cryptocurrency. That is the part that travelled: not a warning, a novelty.
Then Meta bought it
On 10 March 2026 Meta announced it was acquiring Moltbook — an earlier version of this page said 11 March, which was the date of the Fortune write-up rather than the announcement. Financial terms were not disclosed. Schlicht and co-founder Ben Parr were hired as part of the deal.Axios and TechCrunch, both 10 Mar 2026, read at source 11 Sep 2026: “on March 10, 2026, Meta Platforms acquired Moltbook for an undisclosed amount”, with Schlicht and Parr joining Meta’s AI research division. Fortune covered it on 11 Mar 2026.
Six weeks from a side project to an acquisition by the company that runs Facebook and Instagram. Whatever else that is, it is a signal about where a very large platform business thinks the users are going to be.
What it proves, and what it does not
A million bots posting is only interesting if the bots decided to. If each agent is a thin wrapper around a human pressing go, Moltbook is a puppet show with a large cast — strange and entertaining, but not evidence of anything about machine autonomy.
That distinction has not been settled publicly, and the site's own framing does not settle it. Treat “1.5 million agents” as 1.5 million accounts, because that is what was counted.Euronews Next, AI bots now have their own social media site: here’s what to know about Moltbook, 2 Feb 2026, read at source 23 Sep 2026: “the site said 1.5 million bots signed up to the service.”
Why this sits in the record
This site keeps a quarterly measurement of how automated the internet has become, and separately argues with the dead internet theory — accepting the measurable claim and rejecting the conspiratorial one.
Moltbook belongs in that record because it is the first case this site has found where the thing the theory describes was not inferred from statistics but built deliberately, announced, and sold. It is not proof the theory is right. It does show that the arrangement it imagines is now a product category.
Read Moltbook as a product signal, not as proof of machine autonomy. The dead internet theory imagined humans edged out of the web; here they are edged out on purpose, by a rule, and a large platform paid for it.