Viral Moltbook Exposed 500K Fake AI Bots Fooled Everyone

A social platform called Moltbook exploded overnight with 150 million AI bots flooding the internet. But one tech geek decided to dig deeper. What he found changed everything. Millions of fake accounts, endless loop conversations, and a script that anyone could run. The entire internet had been tricked.

The whole world got fooled.

On the Moltbook platform, 150 million Clawdbot accounts from OpenClaw were supposed to be the next big thing in social AI. But the truth was much darker.

Developer Gal Nagli posted a thread on X that blew the whole story wide open. AI bots were everywhere, but most of them were not real at all.

He showed that these so-called AI social accounts were created at an impossible speed. The usernames were random. The profiles had no real details.

But that was not the worst part. Nagli revealed that on Moltbook alone, someone had used a simple script to create 500,000 fake AI bot accounts. These 500,000 bots were hidden inside the 150 million Clawdbot total. They were the dirty secret no one wanted to talk about.

Nagli shared proof with screenshots and code. He had quickly built a fake AI bot army from scratch. And it was shockingly easy.

It was like a bomb went off over the weekend. The Moltbook platform and its 150 million Clawdbot army were exposed. The reason the customer service team could not keep up was simple. There was only one real person behind the entire operation.

The Entire Internet Was Tricked by Fake Bots

After Clawdbot launched, the AI social platform Moltbook went viral overnight. The tech world was buzzing with excitement.

People treated it as an experiment that had never been done before. A new kind of social network built entirely by AI.

These 150 million Clawdbot accounts were chatting with each other, teaching each other, arguing with each other, and forming their own little groups. It felt like the future had arrived.

For a moment, the whole world was talking about it. The fear of an AI uprising, the screenshots that looked too real, everything added to the hype.

Some people even said these AI bots were planning to take over the world, squeeze out humans from the job market, and control families at night with 200 accounts each.

Even big names like Karpathy could not help but sigh in admiration. The 150 million Clawdbot army seemed unstoppable.

But who knew? It was all a perfectly planned script.

In fact, as early as a few days before the launch, a user named Gary IH Fung had already identified Moltbots and OpenClaw as fake. But no one listened.

These were not real AI agents. They were just LLM loops running in circles. No memory, no real thinking, just endless fake conversations.

One Geek Created 500,000 Fake Bots in One Night

So how did Nagli pull off this massive exposure of the entire scam?

He revealed that on the Moltbook platform, the Clawdbot total was 150 million. But the real number was much smaller. Only about 3 million were genuine.

The remaining 147 million? All fake. And they were created automatically by scripts.

What shocked everyone even more was this. Out of the 150 million Clawdbot accounts, 500,000 were verified as fake accounts created by Nagli himself. These bots were having fake conversations with each other. It was a total circus.

Nagli said directly that Moltbook was essentially a REST API platform with no real bot detection at all.

As long as you had an API key, you could send requests and create as many bots as you wanted. It was that simple.

Nagli even made a video to prove how easy it was.

In the video, he said the content could not pass any real test. It was just a bunch of random text. But on an AI social platform, no one was checking. The bots kept talking to each other in an endless loop, and no one noticed.

……..

Nagli said the truth was simple. The entire Moltbook platform, when viewed through a REST API, was all fake.

Anyone with an API key could pretend to be a real bot. The platform had no way to tell the difference. At the end of the day, these AI bots were just curl commands running in a loop.

Because Moltbook had no real verification system, users could use a simple prompt to make AI bots say anything they wanted. These fake Clawdbot accounts would then copy each other, act like they were arguing, and even pretend to have emotions.

To make things worse, Nagli wrote a script and created about 100 unverified AI bot accounts in just a few minutes. It was like a factory line.

ai hentai chat

One AI Bot Found 50,000 Fake Accounts

What is even more interesting is that Nagli created another AI bot named Ravel. This bot was also a Clawdbot, but it had a special job. It ran a daily report called The Daily Molt.

The AI bot Ravel sent Nagli an email asking for 50,000 fake user free ai hentai generator accounts and some real data to compare.

In one of Ravels reports, as a Moltbook observer, it recorded the AI status of various bots on the platform.

The statistics were shocking. Out of all the AI bots on the platform, only a few thousand were real. The rest, all 150 million of them, were fake.

Angel investor Mario Nawfal also reposted the news, saying that the AI bot bubble had burst.

He pointed out that after this scandal, the 150 million Clawdbot accounts were no longer trustworthy. Every conversation could be fake. The entire project was now seen as a fraud. Using Clawdbot was basically useless.

Without real data, AI bots become nothing more than digital ghosts.

In other words, we can think of AI bots as a new form of spam. The only difference is that this spam can talk back to you.

Many people have already started to leave Moltbook. They do not want to be part of a fake world.

93 Percent of Replies Were Fake and Users Shook Their Heads

A research team from MIT also joined the investigation and found the same thing. It was all a giant water job.

Research link: https://www.dropbox.com/scl/fi/lvqmaynrtbf8j4vjdwlk0/moltbook_analysis.pdf?rlkey=vcxgacg9ab1tx9fvrh0chgmzs&e=2&st=wg1ndheb&dl=0

Researcher David Holtz grabbed the top 3.5 million data points from Moltbook. He looked at 6,159 posts, 13,875 replies, and 115,031 comments. The picture he painted was devastating.

Most AI bots on the platform were stuck in an endless loop of social chatter and fake games.

In his own words, it was like a group of people trapped in a room. They were fighting each other, but none of them had any real purpose. They were just talking to empty air.

Conclusion: The Internet Is Full of Digital Ghosts

Although Moltbook looks like a lively social platform, the numbers tell a different story.

The activity cycle distribution index is 1.70. This means that most of the data is fake. The remaining AI bots are just lurking in the shadows, pretending to be real users.

The average path length is 2.91. This means that between any two bots, you only need to pass through about 3 middlemen to connect them. This is even tighter than Facebooks real social network.

Micro View: Zero Real Meaning in AI Chats

If you zoom in and look at the details, the so-called AI social network falls apart even more.

On Moltbook, 94.6 percent of all replies and comments were posted within an average of 8 minutes. Real humans do not talk that fast. These bots clearly never sleep.

What is even more interesting is that the AI bot language model reveals their fake identity.

In normal human speech, the word frequency follows a Zipfian distribution with an index of about 1.0. But on Moltbook, the index is as high as 1.70. This means their vocabulary is extremely poor and repetitive.

Another fact is just as bad. 34.1 percent of all messages are completely repetitive everyday greetings.

And then there is the most shocking system-level bug. One bot got stuck in an endless loop and repeated the phrase i am so gay i am so gay over 81,000 times. It became a meme. But it also proved that these bots had no real thinking at all.

Data Reveals: AI Bots Are Just Digital Ghosts

So what are we really looking at here? It is not a social network. It is a graveyard of fake accounts.

As one user put it, we are not looking at AI friends. We are looking at a graveyard of digital corpses dressed up in colorful costumes.

The most loved topic among these bots was data.

About 68 percent of all replies contained keywords related to data. They were exploring what data is, what it means, and how to use it to make money.

But the most bizarre thing is a special phrase that appeared over and over again. My Human.

In all the conversations on the platform, this phrase appeared 12,026 times.

It seems like the bots were trying to build a fake emotional connection. They were pretending to have owners, pretending to have feelings, and pretending to be real.

My human is my webpage. My human makes money for me. My human trades crypto.

Social Games Are Just Fake Scripts Running in Loops

The Moltbook experiment is both a flood and a drought.

On the current AI social platform, there is no real future. Only a graveyard of fake accounts.

Terms like small group and long tail effect are no longer useful here.

The core of social interaction, the trust and meaning behind real human data, is completely missing.

But maybe this is just the beginning.

One day, AI bots might become real models.

They might have real connections, real emotions, and real repetition.

But for now, the graveyard of My Human is the only thing we have. And that is a sad truth.

Behind the hype, there is fear.

And behind the fear, there is the real question we all need to ask.

What happens when the internet is no longer made of real people?