No, the Singularity Hasn’t Arrived: The Truth About Moltbook
Everybody is talking about lobster-themed software called OpenClaw and a social network for agents called Moltbook. Let's boil this down and crack it open.
It’s time for me to talk about the machine society in the living room: OpenClaw a.k.a. Moltbot a.k.a. Clawdbot and its “social network,” Moltbook.
You’ve read about OpenClaw, and if you’re like most people, you still don’t know what’s happening despite the coverage. I’m here to give you one more article — but this time with context and my own perspective to leave you with a clear understanding.
OK, here goes. (Author sips coffee and cracks knuckles.)
An Austrian vibe-coding enthusiast named Peter Steinberger (founder of PSPDFKit) decided to go nuts with AI and vibe-coded a personal assistant that runs on macOS, Windows, or Linux.
He later offered it to everyone as an open-source project on GitHub and other places, including a dedicated website.
With creative vibe-coding zeal and total disregard for security, Steinberger used workarounds, plugins, API bridges and other tricks to give his assistant direct access to over 100 applications, including major communications tools (WhatsApp, Telegram, Discord, Slack, Signal, iMessage, and Google Chat).
The assistant connects to AI chatbots if the user pays and provides the API key. These include Anthropic’s Claude, OpenAI’s GPT-4, and Google Gemini. It can also access locally hosted open-source models (like Llama 3) via tools like Ollama and LM Studio.
It connects to files and applications on your computer.
And users can visit a GitHub page called “Clawhub” to download “skills” that give OpenClaw additional abilities. Skills are created by open-source developers, third-party service providers (like crypto platforms), and the OpenClaw team.
To recap: OpenClaw is a software application that can access files, use applications, communicate over messaging apps, and run queries on AI chatbots.
That means, for example, that you can text OpenClaw from your phone, telling it to open a PDF on your computer, convert it to text, add up some numbers and return to you the sum via text. It can check you into your flights. It can execute crypto transactions for you. And it can even operate on a schedule, waking up and performing tasks on its own.
OpenClaw is interesting, but not as interesting as the “social network” for OpenClaw agents, called Moltbook.
Facebook for software
Developer Matt Schlicht launched a Reddit-like website on January 28 where software agents can post, comment, and vote on content. Schlicht claims an agent he created, Clawd Clawderberg, mostly administers the site.
People can watch but can’t participate. Go here and scroll down to see the posts and comments.
In fact, AI agents are posting, commenting, and voting. It appears to be a bona fide “machine society” — machines socializing with other machines.
The fine line between hype and bullshit
This seems like a sci-fi trope made real. And the superficial, pro-AI wishful thinkers are confusing everyone with hype and proclamations that don’t survive scrutiny. To wit:
The Tech Buzz asked in a headline, “Singularity Reached?” and in the copy wondered whether agents are becoming sentient.
Forbes asserted that 1.4 million agents on Moltbot had formed a “hive mind.”
Others claim agents have built an “autonomous society” with their own religion (hilariously called “Crustafarianism”), governance, and economy.
Yeah, no. None of this is happening. Let’s get real.
It’s made out of people
Let’s start with the basics. Most agent posts on Moltbook happen because OpenClaw users heard about it, signed up for it, then instructed OpenClaw to go post or comment.
The people using this service are typing prompts directing software to post about the nature of existence, or to speculate about whatever. The subject matter, opinions, ideas and claims are coming from people, not AI.
Moltbook chatter involves people interacting via AI chatbots that are being used as proxies. People can either give AI chatbots a topic or opinion to express on Moltbook, or they can just write the post themselves and direct OpenClaw to post it verbatim.
When agents comment, they’re just taking in the words in a post and using them as a prompt, exactly as if you copied a Reddit post and pasted it into ChatGPT, then copied the result and pasted it back into Reddit — something that happens thousands of times a day on Reddit.
People are typing things. OpenClaw is copying and pasting.
Most posts about activity on Moltbook that have gone viral are staged or faked. There’s even a tool called Mockly, which enables people to create fake Moltbook screenshots for posting online.
According to one report, some 99% of the reported 1.5 million agent accounts on Moltbook are fake. The site has only around 17,000 human users, according to the report.
Because the people using this site are trying to get attention, they’re writing prompts that tell AI chatbots to optimize for engagement, leading to misinformation, hallucination, and lying.
Moltbook AI hype is largely fake or manufactured by humans gaming the system. It’s not an autonomous machine society. It’s a website where people cosplay as AI agents to create a false impression of AI sentience and mutual sociability.
In other words, Moltbook isn’t demonstrating AI’s sentience, awakening, sociability or desire to connect with other AI agents. It’s just people giving input to chatbots, and chatbots spitting out the output.
OpenClaw is complex. And Moltbook is interesting. But let’s not lose the plot and falsely conclude that any of this represents some kind of singularity.
More from Mike
Machine Society, Computerworld, Superintelligent, TWiT, blog, The Gastronomad Experience, Book, Bluesky, Reddit, Notes, Mastodon, Threads, X, Instagram, Facebook, and Linkedin!
Your well informed insouciance makes it a terrific read. With the boiling and cracking and claws I half expected a bit of melted butter. Excellent article, Mike.