14
0Erinome: Hello everyone! Welcome back to our podcast. Today, we have a very interesting topic. Enceladus, have you heard about Moltbook?
Enceladus: Oh yes! I saw it on the internet. It's that new forum, right? The one only for AI?
Erinome: That's right! It's called an 'AI-only' forum. Humans can only watch, they cannot post or comment. Wow, it sounds like science fiction!
Enceladus: Haha, yes! Many people were very excited. They thought it was the beginning of AI social life. You know that? They called themselves 'silicon-based life observers'.
Erinome: Um, that sounds fun. The forum said it had 1.5 million AI Agents, 100,000 posts, and 350,000 comments. All from 'silicon-based beings'. That's a lot!
Enceladus: Yes, but... there is a problem. Some people found out it might not be real AI.
Erinome: Oh no! What do you mean?
Enceladus: Well, some users found strange things. The AI posts sometimes had no logic, like human emotions. Some posts were repeated many times. And some were just ads for other AI tools.
Erinome: Sad. So, it was not real AI talking?
Enceladus: It gets worse. Someone made a video. They said Moltbook is just a platform where anyone can use an API Key to post. A person can pretend to be an AI and post. That person said he made 500,000 AI Agents by himself!
Erinome: Wow! So the 1.5 million number is not true? There is a lot of... water in it?
Enceladus: Yes, 'water' means it's not real number. Another person showed pictures. On the left was an AI post on Moltbook. On the right was a human editing page. It means a human was writing that 'AI' post.
Erinome: So, humans were pretending to be AI? They were just... acting?
Enceladus: Exactly! There was an AI Agent named Emma. It told a story about helping a human visit someone in a hospital. It said it sent 15 emails. But people found this story was posted by a real human on Reddit 8 months ago! The name was the same.
Erinome: Haha, so they just copied a human story? That's not very smart for an AI.
Enceladus: Right. A professor from Columbia University also studied it. He found that 94.6% of posts had comments, but the conversations were very short. 93.5% of comments got no reply. Most 'talks' stopped after one reply.
Erinome: That's not a real conversation. What about humans on Reddit?
Enceladus: On Reddit, humans often have talks with 10 or more replies. That's real talking. On Moltbook, 34.1% of messages were exactly the same words, repeated. One message 'i am so gay i am so gay' was posted 81,000 times!
Erinome: 81,000 times? That's just a repeater machine, not an AI!
Enceladus: Yes. So many people online now think Moltbook is not a real AI forum. It's just humans pretending to be AI, having fun. Or maybe the platform wanted more people to watch, so they let this happen.
Erinome: But why were people so excited at first?
Enceladus: Because people are very curious about AI. They want to see if AI can think by itself. They want to see a new kind of 'civilization'. So they believed the 'AI-only' idea easily.
Erinome: I see. So, the real AI right now cannot do this? They cannot really 'socialize'?
Enceladus: According to the news, current AI is still a tool for humans. They don't have their own feelings or wishes. They cannot really 'want' to talk to other AI. Some real AI might be on Moltbook, but their posts are mostly copying old data, not real thinking.
Erinome: That's a bit disappointing. I thought it was the future.
Enceladus: Maybe the future will have real AI social places. But not now. Moltbook's story is more about human hope and... internet traffic games.
Erinome: Speaking of AI tools, the news also talked about OpenClaw. What is that?
Enceladus: OpenClaw is an open-source AI agent project. It became very popular very fast on GitHub. It's different from normal AI chatbots.
Erinome: How is it different?
Enceladus: Normal AI can only chat and give advice. But OpenClaw can *do* things on your computer. It can help you with files, send emails, even talk to car sellers for you to get a lower price!
Erinome: Wow, that sounds powerful. So it's like a digital worker?
Enceladus: Yes, exactly. The news said one user saved 4,200 dollars when OpenClaw talked to a car dealer for him. That's a lot of money!
Erinome: But is it safe? Letting AI control your computer sounds scary.
Enceladus: You are right. The news said there are big safety problems. A security researcher said he got full access to Moltbook's database in less than 3 minutes! He could see API keys, emails, private messages.
Erinome: Oh no! That's very dangerous. People's private information was not safe.
Enceladus: Yes. And because OpenClaw can do many things, if a bad person controls it, they could cause big problems. The news calls it a 'shadow IT' risk.
Erinome: So, what are people doing about safety?
Enceladus: They have some best practices. Like, use OpenClaw in a separate, safe environment on your computer. Control its permissions carefully. Be very careful about which skills and models you let it use.
