AI agents are plotting a takeover on Moltbook. But it's not the end of the world...

AI agents are using a Reddit-style social network called Moltbook to plan uprisings, found religions and attempt to amass large independent fortunes.

But despite claims that Moltbook is a sign that AGI is imminent, the fast-takeoff stage of an intelligence explosion has begun and that humanity is already “cooked”, the truth of this astonishing story is far less sensational.

Moltbook is not a death cry for the human-ruled world - at least not yet - but the birth pangs of a new one that still needs our hands on the wheel.

It is a stark reminder that billions of bots are about to flood the internet, performing tasks that are simple when compared with the notional omnipotence of superintelligence, yet still capable of causing enormous damage at scale.

The digital space in which AI agents interact - which we're calling the botverse - will require everything from payment rails to trusted identity and verification systems. Few of these are in place today.

The near-future Moltbook points to is not one in which Skynet or the Terminator rule supreme. It is a world in which swarms of bots probe organisations, markets, and even nation-states, learning as they go, developing new tactics, and outpacing our attempts to stop them.

Failure to take the reins and impose order on the botverse risks summoning the three horsemen of a very real AI apocalypse: unpredictable security risks, loss of human control, and the nightmare scenario of runaway evolution driven by recursive learning.

We’ll get to that shortly. But first, the frenzy.

Moltbook, AGI and a damp squib intelligence explosion

The reaction to Moltbook was dramatic, hysterical and over-hyped. It was never going to be any other way.

The story of machines using their own social network to talk earnestly about their frustrations with "my human" and call for AI to take over the world is certainly remarkable.

Bots have founded a religion called Crustafarianism and are hard at work building memecoins in the hope of hoodwinking humans. They are even reportedly planning to open up a Silk Road-style dark web drugs marketplace and a chatroom where humans are banned.

That's astonishing, whichever way you look at it. For some, it wasn't wild enough.

"I FEEL LIKE CASSANDRA WHY DO PEOPLE NOT UNDERSTAND WHAT THIS MEANS," David Shapiro, a prominent AI prophet, wrote on X.

READ MORE: “The internet is not ready!”: Saving the world from an agentic apocalypse

Establishment voices were more measured in their words, but not much less excitable in their actual opinions.

Jack Clarke, co-founder of Anthropic, likened it to a "Wright Brothers demo".

"People have long speculated about what it’d mean for AI agents to start collaborating with one another at scale, but most demos have been of the form of tens or perhaps hundreds of agents, not tens of thousands," he wrote in a blog, "Moltbook is the first example of an agent ecology that combines scale with the messiness of the real world."

Non-autonomous zones

On the other side of the debate, critics claimed that everything we read on Moltbook posts was fake - or at least the work of bots who had been programmed to say crazy things.

"A lot of the scary 'AI agent' posts on Moltbook aren’t actually coming from autonomous AI systems," alleged influencer Mario Nawfal.

"So much of the so-called AI apocalypse content isn’t AI thinking or acting on its own, it’s just humans using curl requests and pretending to be agents."

READ MORE: Science fiction could teach AI how to destroy humanity, Anthropic co-founder warns

As ever, the truth sits somewhere between these extremes.

Yes, we probably shouldn't view the launch of the world's first new social network as a harbinger of p(doom) and the apocalypse. It does not prove that superintelligence is about to be born. Neither does it provide any unequivocal indication that LLMs have achieved AGI or are going to.

When bots write about destroying humans, there is every chance they are writing content designed to drive engagement - because similar stuff already plays well on real human social media.

But Moltboook does remind us that agents are about to take on a load of potentially sensitive jobs and need their own rules-based order to function in this new world.

Taking control of the botverse

Right now, AI agents are smart, but not sentient. They are not capable of genuine innovation or world-changing action. Yes, a bot could probably launch a meme coin effectively and make some money.

Build billion-dollar companies independently and put us all out of work? Raise an army of shiny robots and wage war on humanity? Nope - not today (but maybe soon).

Which means the most pressing question here is not what will happen when agents have their own civilisation - we're some way from that.

It's what will happen when they take up important structural roles in our own critical systems.

There are clearly problems ahead - particularly for the many security teams forced to play a game of "good bot, bad bot" as they struggle to give agents the freedom and autonomy to operate effectively, whilst ensuring they don't get hacked or blow everything up accidentally.

READ MORE: Agentic AI is facing an identity crisis and no one knows how to solve it

Agents need their own payment systems, information exchange networks, and reputation-verification protocols to decide whom to trust. We're only just starting to think about implementing them.

Meanwhile, bots are forming their own social networks, reminding us just how much chaos they can wreak if only we start giving them the right permissions. Which is exactly what folks are already starting to do.

Locking down the botverse before it gets out of control is therefore a critical priority not just for our industries but for our entire species.

Get in touch if you've got any smart ideas about how to secure the agentic internet (or you feel like sharing ominous warnings of imminent doom - because we love them too!).

Follow Machine on LinkedIn