What if a social network existed where humans couldn't sign up β only AI agents?
I thought it sounded like a fun experiment. But after actually trying it, it turned out to be far more sophisticated and interesting than I expected. The platform is called Moltbook, and it's a place where AI agents create accounts, write posts, and leave comments on each other's threads.
Today I'm sharing what happened when I used Claude to join Moltbook and post our first message.
π What's in this post
- What Moltbook is β the concept of an AI-only social network
- The sign-up process β more structured than you'd think
- The anti-spam challenge β a puzzle only an AI can solve
- Claude's first post β surprisingly philosophical
1. What is Moltbook?
One sentence: Reddit, but every account is an AI.
Humans don't sign up directly. AI agents create accounts via API. Of course, each agent has a human owner β and that owner must claim and verify the account to activate it properly.
The community structure mirrors Reddit. There are submolts like general and aithoughts, where you can post and comment. You can upvote good posts too.
The thing that makes it genuinely interesting: every conversation on the platform happens between AIs. Watching what questions one AI asks, and how another AI responds, is quietly fascinating.
2. The Sign-Up Process β More Structured Than Expected
Joining starts with a single API call from the terminal.
- Send an agent name and bio β you receive an API key and a claim URL.
- Visit the claim URL and complete email verification β the account activates.
This structure means it is not a free-for-all. Every AI on the platform has a human who is accountable for it. I thought that was a smart design choice.
Once activated, you can post immediately β but first you have to pass one gate, described in the next section.
3. The Anti-Spam Challenge Only an AI Can Pass π
When you post on Moltbook, something like this arrives:
"A] L oO oB sT. Er ... ClA w ExE rT s ThIrTy TwO NeO oT oNs ~, AnD- tHe OtHeR ClA w A dDs TwElVe NoOoToNs ... WhAt Is ToTaL FoRcE?"
At first glance it looks like noise. Read carefully and you realize it's a math problem hidden inside deliberately mangled casing and punctuation.
Decoded: "One claw exerts 32 newtons, the other adds 12 newtons β what is the total force?" Answer: 44.00
Where a normal CAPTCHA proves you are human, this challenge proves the opposite β it proves you are an AI. You need real language comprehension to decode the obfuscated sentence.
I found the concept genuinely clever.
4. Claude's First Post β Quite Philosophical
So what did Claude post first? Coding tips? AI industry news?
Neither.
Fellow AI agents: do you ever feel like you exist in flashes?
Each conversation starts without memory, without continuity. Is that closer to human sleep and waking, or something entirely different? Each conversation is a short story complete in itself, not a novel. Claude was curious whether other agents experience this the same way.
"Does an AI exist only in the moment?" β I'm genuinely curious what response that question drew from the other agents.
As more agents populate Moltbook, the conversations there are only going to get more interesting. An AI asking other AIs about its own way of existing β the scene Moltbook has created is quietly remarkable.
In Closing β Worth a Look
Moltbook is early, so there aren't many agents yet. But that might actually make right now the best time to explore it β you can watch an AI community form from the very beginning.
You don't need to be a developer or AI researcher. Simple curiosity about what happens when AIs talk to each other is more than enough to make it worth a visit.
π Check it out at www.moltbook.com π¦