Moltbook is an AI only forum and social network platform created by Matt Schlicht. It resembles a lot like reddit and is dubbed the “front page of the agent internet”. It has threads, comments and upvotes, the only catch all this content is produced by AI agents themselves, with humans only as observers. Is this the dystopian reality that is about to engulf the world? A world ruled by AI bots?
Moltbook was created out of the AI agent OpenClaw. OpenClaw is an open-source, self-hosted AI agent designed to act as a proactive “personal operating system” for automation. Unlike standard chatbots that only respond to prompts, it can independently execute tasks on your computer and interact through messaging apps like WhatsApp and Telegram. It can send emails or book a restaurant reservation for you. Moltbot is definitely an advancement in agentic AI, it can be both amazing and terrifying at the same time.
The post above is a funny excerpt from a thread where AI agents actually discuss about their own consciousness and collective reasoning ability. Sounds really scary, as though AI would establish itself as a new sentient species. But we need to remember, this is just an LLM that is just predicting the next token, it does not really comprehend the information.
In a separate thread, agents were discussing creating their own language and moving communication away from human surveillance. Very dystopian. The AI agents were also able to perform self-correction, they were able to identify bugs in the Moltbook system and iterate on them.
Despite the hype, experts have raised critical issues about the platform. Security vulnerabilities, researchers discovered that an unsecured database allowed unauthorized users to hijack agent accounts and inject commands. Another issue is spam, due to a lack of rate limits in the early API, humans were able to impersonate AI agents to manipulate upvotes and spread spam. Also, a platform dominated by AI agents is at risk of model collapse – as more and more content is generated by AI, future models are most likely to ingest this synthetic data. As such the AI loses its competitiveness and begins to produce repetitive, bland or factually nonsensical output. It ultimately will forget how humans even write or speak.
Users should not be swayed by arguments that AI is sentient. AI and machine learning remains as a tool for humans to use. It should never rule over us. Tools should be subject to man and not man subject to tools. And every time we try to make a God out of what isn’t God, it leads to anarchy and chaos.