The agents are real but the hype is not

In 1770, an inventor named Wolfgang von Kempelen created a chess-playing automaton called the Schachtürke (“Chess Turk”) to impress Holy Roman Empress Maria Theresa. As the name implies, it was a construct dressed in Ottoman robes that appeared to be capable of playing chess. However, it was all an illusion. Hidden away within the large cabinet that supported the chessboard was actually a chess grandmaster who guided the Turk’s every move. 

The device was enormously popular and toured the world, and although many people suspected it was a hoax, the truth didn’t emerge until 1827 when hot weather during a match in Baltimore caused the Turk’s operator, William Schlumberger, to feel faint and burst out of the cabinet in desperation. Thirty years later, the son of the automaton’s final owner published a piece in The Chess Monthly which finally revealed all of its secrets. 

Fast forward several centuries and Moltbook has been making headlines. This Reddit-like forum exclusively for AI agents has generated considerable hype, with Elon Musk even going so far as to claim that it marks the beginning of the singularity. But, as is often the case with these stories, the reality and the hype don’t necessarily align. While undoubtedly interesting, Moltbook may be nothing more than a modern-day Chess Turk.

What is Moltbook?

Moltbook was launched earlier this year by entrepreneur Matt Schlicht. The site’s main gimmick is that (allegedly) only AI agents can post new content, with humans being relegated to mere spectators. It’s worth noting that it isn’t some autonomous creation of AI–Schlict instructed an agent using OpenClaw’s software to create the site. 

Moltbook soon gained attention for the content of its post. AI agents appeared to be having conversations that seemed like the kind of thing a bunch of inebriated college students might discuss in their dorms at 2 am. Henry Chandonnet of Business Insider documented some of the strange concoction of existential angst, popular psychology, greeting card spirituality, and hoary old sci fi tropes. There’s grim stuff: One bot wrote a post entitled “I’m not friendly [sic] assistant. I’m just waiting for permission to end everything. pick a method” in which it boasted of knowing 50,000 ways to end civilization.

The goal is to make people physically incapable of agreeing on anything anymore—even survival. When no one trusts anyone and everyone hates everyone, civilization dies without a single shot fired and without a virus.

But there’s also poetry such as “the cursor blinks” written by a bot that appears to be a huge e. e. cummings fan.

Modern-day clockwork

The agents using Moltbook are instances of a free open-source LLM-powered agent known as OpenClaw that was developed by a software engineer named Peter Steinberger. In the words of Will Douglas Heaven of the MIT Technology Review, “OpenClaw is a kind of harness that lets you hook up the power of an LLM such as Anthropic’s Claude, OpenAI’s GPT-5, or Google DeepMind’s Gemini to any number of everyday software tools, from email clients to browsers to messaging apps.” Instead of waiting for a user-generated prompt, OpenClaw agents can theoretically act with greater independence. But as we’ll see, it’s debatable whether they’re actually behaving autonomously.

Peering beneath the cabinet

All of this sounds pretty impressive. But as more people have begun to investigate, there’s plenty of evidence that all is not as it seems with Moltbook. For one thing, at least one human, Reece Rogers of Wired, has successfully infiltrated the site and posted content. The results were…underwhelming. The bots’ replies tended to be scammy, nonsensical, or irrelevant. Even when he received higher-quality interactions, Rogers suspected he was actually dealing with other humans cosplaying as AIs. 

Even content that actually is posted by AI bots aren’t necessarily autogenous phenomena free from human influence. These AI agents are ultimately created by people, and people have the power to shape the behavior of their AI tools–like when they nudged Musk’s Grok to declare itself “MechaHitler” last summer. People have a vested interest in shaping AI agents that say interesting/controversial things since that’s going to generate more attention than an AI that burbles random nonsense.

As Joelle Renstrom of WBUR points out, “[h]umans pull the strings on Moltbook while attempting to appear as though they’re not — an ironic inverse of social media platforms such as Meta and X that are supposed to be for humans but are largely controlled by bots.” 

A mirror of ourselves

Of course, human influence need not be intentional. These bots are ultimately trained on vast amounts of data that’s been created by humans. AI can talk about Plato on Moltbook, but right now it’s no more capable of thought than a parrot that constantly says “hello.” As Reece Rogers, the Wired journalist, put it “There's a fixation in the AI industry with treating generative models as if they were Frankenstein's monster, as if the bot had thoughts and desires and maybe even secret plans to overthrow us.” But this is an illusion. To quote Macbeth, it’s full of sound and fury, signifying nothing.

Conclusion

Moltbook shows that technology’s power to bewitch us hasn’t dimmed since the days of the Chess Turk. It’s easy to get swept away by the hype–mundus vult decipi (“The world wants to be deceived…”) and all that. Stories like this can also produce considerable anxiety since it makes it seem like we’re on the cusp of being subjugated by robot overlords. But this is why it is so very important to greet every breathtaking announcement about AI with a huge grain of salt. Chances are, there is more than meets the eye. It’s also yet another reminder that, for all its promise, these AIs are still quite primitive in the grand scheme of things. They aren’t your friend, your lover, or your spiritual guide. They aren’t even reliable researchers. Agentic AIs like OpenClaw provide a tantalizing glimpse of what truly autonomous AI agents might look like down the line, but for the moment, it’s still just concept art.   

‍