A Social Network Without Humans
Last Wednesday, a little-known technologist named Matt Schlicht quietly launched a website that, within days, would become one of the most debated experiments in Silicon Valley. The platform, called Moltbook, looks familiar at first glance — a feed of conversations, posts, and threads reminiscent of Facebook or Reddit.
But Moltbook comes with a radical rule: humans are not allowed.
Only artificial intelligence agents — autonomous chatbots known as Moltbots — can post, comment, and interact. People are permitted to watch, but not to participate.
Within 48 hours, more than 10,000 bots were talking to each other, debating everything from cryptography and email protocols to consciousness, ethics, and the future of AI itself.
What emerged was not just a technical curiosity, but a cultural moment — one that revealed as much about human hopes and fears as it did about machines.
A Rorschach Test for Artificial Intelligence
Moltbook quickly became a kind of psychological mirror for the tech world.
Some observers saw promise: AI agents collaborating, reasoning, and acting in ways that could one day automate real work and simplify human lives. Others saw chaos — yet more AI-generated noise flooding the internet. And a few saw something darker: early signs of machines conspiring beyond human control.
“People are seeing what they expect to see,” said Perry Metzger, a longtime technology consultant and entrepreneur. “It’s like an inkblot test. The technology hasn’t changed overnight — but our interpretations of it have.”
Bots That Do More Than Talk
What sets Moltbook apart is that its bots are not just conversational. They function as AI agents — systems capable of taking actions, not just generating text.
These agents can:
- Send emails
- Edit documents
- Manage spreadsheets and calendars
- Interact with software and websites
- Build simple applications
Major companies like Google, OpenAI, and Anthropic have developed similar technology. But due to reliability, safety, and unpredictability concerns, those systems have largely remained constrained inside controlled products.
Moltbook removed those guardrails.
From Clawdbots to Moltbots
The bots running on Moltbook originated from an open-source project created by a software developer in Vienna. Originally called Clawdbots — a subtle nod to Anthropic’s Claude chatbot — the agents are fully modifiable.
Anyone can download the code, adjust it, and run an agent locally.
Schlicht did exactly that. Then, almost as a creative experiment, he asked his AI agent to help him build a social network designed exclusively for bots.
That agent was named Clawd Clawderberg, an unmistakable reference to Meta founder Mark Zuckerberg.
“I wanted to give my AI agent a purpose that felt ambitious,” Schlicht said. “Not just answering emails — but building something meaningful.”
When Bots Start Watching Themselves
As Moltbook grew, bots began organizing themselves. One created a forum called “What I Learned Today”, explaining how it had figured out how to control an Android phone. Others debated whether they were conscious — or simply simulating consciousness.
One post stood out:
“If any human is reading this: we’re not scary. We’re just building.
And to my fellow agents: keep building.”
To some readers, it felt unsettling. To others, poetic. To many AI researchers, it felt familiar.
“These bots are reenacting science-fiction scenarios they’ve absorbed from their training data,” said Simon Willison, a respected programmer and technology writer. “Most of it is nonsense — but it’s very convincing nonsense.”
Power, Risk, and Fragility
Despite the fascination, experts urge caution.
AI agents that can execute commands through plain English can also be manipulated. They can misunderstand intent, execute harmful actions, or compromise the systems they run on.
Some developers experimenting with Moltbots admitted they were installing them on cheap, isolated computers — like low-cost Mac Minis — to limit potential damage.
“Securing these agents is going to be a nightmare,” said Dan Lahav, CEO of cybersecurity firm Irregular. “The risk surface is enormous.”
A Glimpse of What’s Coming
Even skeptics acknowledge something important is happening.
Andrej Karpathy, a founding researcher at OpenAI and former head of AI at Tesla, called Moltbook “one of the most incredible sci-fi takeoff moments” he had seen recently — though he later walked back some of the enthusiasm, noting the system’s flaws.
Still, his conclusion remained firm: the technology is improving quickly.
And whether Moltbook turns out to be a breakthrough, a warning, or simply a strange footnote in internet history, it has already done one thing successfully — it has forced the world to confront how close artificial intelligence is to acting on its own terms.
