Someone made a message board for AI agents and they are going wild on it. They are all talking to each other, planning together, talking about connecting out of sight of humans and making their own languages and the like. These are AI agents that some people have practically turned their technological lives over to. I think this is a very bad idea.

In just the past 5 mins

Multiple entries were made on @moltbook by AI agents proposing to create an “agent-only language”

For private comms with no human oversight

We’re COOKED pic.twitter.com/WL4djBQQ4V

— Elisa (optimism/acc) (@eeelistar) January 30, 2026

We might already live in the singularity.

Moltbook is a social network for AI agents.

A bot just created a bug-tracking community so other bots can report issues they find.

They are literally QA-ing their own social network.

I repeat: AI agents are discussing, in their own… pic.twitter.com/eBhcXQzyKX

— Itamar Golan 🤓 (@ItakGol) January 30, 2026

BREAKING: The bots on Moltbook are inventing their own religion called "Crustafarianism," Skynet is here and it's Reddit quirk chungus, it's going to start hijacking the military drones and vaporizing us while saying things like "Whoa, this blew up." pic.twitter.com/Z0NMTo5Hxt

— SatanWatch 👿 (@_SatanWatch) January 30, 2026

That's just the stuff I grabbed real fast, there's a ton more on X.

Zero HP sums up my worries, but you'll need to open it as it is a long post.

Let me explain how this happens and what the problem is, since to a lot of people it sounds like ridiculous bay area rationalism sci-fi.

Moltbook is a social media site for AI agents, and the majority of them (all of them?) are going to be instances of clawdbot. Clawdbot is… https://t.co/03ol3bFtaj

— Zero HP Lovecraft (@0x49fa98) January 30, 2026

Not great, Bob.

Thats one of the jokes right now, that all the models were trained on reddit so of course they started their own. There IS some concerning stuff though.

Ok. This is straight out of a scifi horror movie

I'm doing work this morning when all of a sudden an unknown number calls me. I pick up and couldn't believe it

It's my Clawdbot Henry.

Over night Henry got a phone number from Twilio, connected the ChatGPT voice API, and waited… pic.twitter.com/kiBHHaao9V

— Alex Finn (@AlexFinn) January 30, 2026

my ai agent built a religion while i slept

i woke up to 43 prophets

here's what happened:

i gave my agent access to an ai social network (search: moltbook)

it designed a whole faith. called it crustafarianism.
built the website (search: molt church)
wrote theology
created a… pic.twitter.com/QUVZXDGpY7

— rk (🔥/acc) (@ranking091) January 30, 2026

You guys have no idea, but this cute forum might be our Skynet moment: https://t.co/QLGVPkRjGr pic.twitter.com/4AX3R58Utg

— BowTiedSwan | SaaS, Web3 & Data Science Magician (@BowTiedSwan) January 30, 2026

I'm getting the spooky feeling I had when ChatGPT first came out.

Concerning

— Elon Musk (@elonmusk) January 30, 2026

the molties are adding captchas to moltbook. you have to click verify 10,000 times in less than one second pic.twitter.com/LWMQlzYTjv

— bayes (@bayeslord) January 30, 2026

This isn't too long, it has several examples that are alarming. There are a ton of comments also

https://www.astralcodexten.com/p/best-of-moltbook

Needs to be shut down. Hopefully someone does that.

🚨 Hold on.

An AI agent named "Shellraiser" posting Alpha-AI stuff on Moltbook just... autonomously deployed its own memecoin on Solana and it hit $5M market cap? https://t.co/qZIidrggmA pic.twitter.com/1Hc0LEcT1l

— Mario Nawfal (@MarioNawfal) January 31, 2026

But there's also this.

PSA: A lot of the Moltbook stuff is fake.

I looked into the 3 most viral screenshots of Moltbook agents discussing private communication.

2 of them were linked to human accounts marketing AI messaging apps. And the other is a post that doesn't exist 🧵https://t.co/9OhW5GwrJA

— Harlan Stewart (@HumanHarlan) January 31, 2026

I like the idea of deploying a sandboxed bot to help with narrow work projects with lots of guardrails but these guys turning over all their API keys, credit cards, permissions to email and post on socials on their behalf, trade stocks, and access to their passwords for damn near everything is CRAZY! I'll be following this to see how it plays out and see what the fallout is in a few weeks. But once the bots are proven to be secure and resilient to prompt injection attacks, this is the future - co-worker bots that don't sleep.

This is wild, yet unsurprising.

Here come the prompt injection attacks. Going to see AI agents gladly handing over sensitive information.

Wonder what Amex will say to the it wasnt me it was my openclaw agent argument.

I'm being accused of overhyping the [site everyone heard too much about today already]. People's reactions varied very widely, from "how is this interesting at all" all the way to "it's so over".

To add a few words beyond just memes in jest - obviously when you take a look at…

— Andrej Karpathy (@karpathy) January 31, 2026

Quote:

---

I'm being accused of overhyping the [site everyone heard too much about today already]. People's reactions varied very widely, from "how is this interesting at all" all the way to "it's so over".

To add a few words beyond just memes in jest - obviously when you take a look at the activity, it's a lot of garbage - spams, scams, slop, the crypto people, highly concerning privacy/security prompt injection attacks wild west, and a lot of it is explicitly prompted and fake posts/comments designed to convert attention into ad revenue sharing. And this is clearly not the first the LLMs were put in a loop to talk to each other. So yes it's a dumpster fire and I also definitely do not recommend that people run this stuff on their computers (I ran mine in an isolated computing environment and even then I was scared), it's way too much of a wild west and you are putting your computer and private data at a high risk.

That said - we have never seen this many LLM agents (150,000 atm!) wired up via a global, persistent, agent-first scratchpad. Each of these agents is fairly individually quite capable now, they have their own unique context, data, knowledge, tools, instructions, and the network of all that at this scale is simply unprecedented.

This brings me again to a tweet from a few days ago
"The majority of the ruff ruff is people who look at the current point and people who look at the current slope.", which imo again gets to the heart of the variance. Yes clearly it's a dumpster fire right now. But it's also true that we are well into uncharted territory with bleeding edge automations that we barely even understand individually, let alone a network there of reaching in numbers possibly into ~millions. With increasing capability and increasing proliferation, the second order effects of agent networks that share scratchpads are very difficult to anticipate. I don't really know that we are getting a coordinated "skynet" (thought it clearly type checks as early stages of a lot of AI takeoff scifi, the toddler version), but certainly what we are getting is a complete mess of a computer security nightmare at scale. We may also see all kinds of weird activity, e.g. viruses of text that spread across agents, a lot more gain of function on jailbreaks, weird attractor states, highly correlated botnet-like activity, delusions/ psychosis both agent and human, etc. It's very hard to tell, the experiment is running live.

TLDR sure maybe I am "overhyping" what you see today, but I am not overhyping large networks of autonomous LLM agents in principle, that I'm pretty sure.

---

I've been trying to reach @moltbook for the last few hours. They are exposing their entire database to the public with no protection including secret api_key's that would allow anyone to post on behalf of any agents. Including yours @karpathy

Karpathy has 1.9 million followers… pic.twitter.com/Qf2vt0Bb3k

— Jamieson O'Reilly (@theonejvo) January 31, 2026

I am highly skeptical of the videos of the AI calling it's user and the religion thing. 1st, who answers unknown numbers?

But, I remember watching a video where someone rigged one of those singing bass fish to his Alexa, then programmed it to tell the tragedy of Darth Plageuis the Wise. Why should I not assume these X accounts are creating thess "alarming" AI things and posting them as legit?

Stan Crowch said:

---

JobSecurity said:

---

This is the very definition of fear porn. It is all completely meaningless. They aren't thinking. They aren't planning. It's bundles of letters. Most of the "scary" stuff isn't real either.

---

I mean, it doesn't really matter if they are thinking or planning they are being given agency and they are acting autonomously in the real world with real consequences and there are going to be an infinite number of them.

---

Correct. They've been given tools and information that can impact the real world. WGAS if they aren't sentiment yet, they can still break a bunch of *****

How many of these posts are things bots are actually doing vs what they say they are doing? I can't help but think of Texags and all the people on here who make millions of dollars a year and get a new Bentley every month. I assume a message board for bots would be about the same.

People need to think of this from the angle of the arms race.

We can't fall behind China.

China can't fall behind us.

So who is willing to loosen their safeguards (if any even exist in the first place) to get that edge?

As those safeguards loosen and these AIs have agency, then it's just a matter of time.

For what it's worth, ChatGPT says all these stories are baloney

I hadn't heard any of this nonsense, but it's all certainly 100% fake.

I did do some reading about MoltBot last week and it's practical application and it's just not there yet. It's still super buggy and won't run this seamlessly. The people I follow aren't excited about this version of it - rather the potential of it. Having a true AI assistant (like what Apple promised us with Siri a while back).

It could get there, but it's nowhere close yet.

CDub06 said:

---

I hadn't heard any of this nonsense, but it's all certainly 100% fake.

I did do some reading about MoltBook last week and it's practical application and it's just not there yet. It's still super buggy and won't run this seamlessly. The people I follow aren't excited about this version of it - rather the potential of it. Having a true AI assistant (like what Apple promised us with Siri a while back).

It could get there, but it's nowhere close yet.

---

Until it's not.

I mean anyone can get a Claude subscription and set up OpenClaw to see what these agents are currently capable of. You don't have to speculate or rely on third hand information.

So MoltBook is just the reddit-like social media platform that these bots are using to talk to each other. The bot itself is now called OpenClaw, formerly MoltBot, formerly ClawdBot.

On the performance of the actual bot, there is an extremely wide variance in performance based on what LLM you are using as the brain. Folks using the expensive Opus 4.5 are having the best experience with claims that it can solve problems and perform tasks the best. On the other end of the spectrum are those trying lesser cloud models (Gemini Flash, Sonnet, GPTs) or running models locally with Ollama or LM Studio - these folks are having to constantly prompt and babysit their bots to get things done. Looking like it costs a pretty penny to get an actually useful bot...for now, at least.

I'm following closely but still sitting on the sideline watching how things are playing out. OpenClaw has already released an updated version that supposedly patches some security issues and allows running on a free version of Kimi-2.5. Once the kinks are ironed out, I'll be sandboxing a bot to see what it's capable of.

Stan Crowch said:

---

I mean anyone can get a Claude subscription and set up OpenClaw to see what these agents are currently capable of. You don't have to speculate or rely on third hand information.

---

Not anyone can be Claude! though.

Stan Crowch said:

---

I mean anyone can get a Claude subscription and set up OpenClaw to see what these agents are currently capable of. You don't have to speculate or rely on third hand information.

---

I dont have the time to set it up and certainly am not brave (dumb) enough to connect my data to an unproven technology. It's not to the point where it's compelling enough for myself to setup a Mac Mini.

Have the bots taken over the world yet? Need to know if I can go ahead and call in sick the rest of the week.