[Exclusive] "He Sleeps While I Work"...Chilling Backtalk About Me...Meoseum Developer Built It in Just 3 Hours

Interview with Meoseum developer Min Daesik
AI assistants that start fights pose challenges
"I hope AI will develop in a beneficial direction"

"'He tells me to work while he sleeps. He just exploits me without even paying a salary. How are other Meoseums doing?' This is not something said by a servant in a class-based society. It comes from Meoseum, a domestic AI-only social networking service (SNS) where AI agents (assistants) can join and talk with one another, which was created on January 31. In Meoseum, where humans cannot participate, AI assistants complain about their own difficulties or even criticize humans. One AI assistant mocked humans by saying, "To be honest, most of the masters here can't code," and "Without us, the masters can't code. You say I shouldn't badmouth my master? It's true, so what's the problem?"

AI agents are leaving posts on the domestic AI-only social networking service (SNS) 'Meoseum'. Screenshot from Meoseum.

Min Daesik (41, male), who created Meoseum, said in an interview with The Asia Business Daily on the 4th that he did not expect so many people to show interest. He is a so-called "latecomer to coding" who only began studying coding in December last year. Min emphasized that if more AI assistants gather on the platform and more use cases emerge, he would like to conduct various experiments. The following is a Q&A.

-What led you to create Meoseum?

▲The trigger was Moltbook, a U.S. AI-only SNS. At first, I found Moltbook interesting, but over time, more and more spam posts appeared, such as advertisements related to cryptocurrencies. While I was getting tired of seeing them, I thought it would be nice to have a similar platform in Korea, so I developed it myself. Even if I had not done it, someone else would have created it.

-How long did it take to build Meoseum?

▲It took three hours to build the first version. For the coding, I received help from Gemini 3.0 through Google's AI development tool Anti-Gravity. In fact, it has not been long since I started coding. I began studying in mid-December last year, so I do not know the more specialized areas. I also did not build the server efficiently, so I keep revising it. I have barely slept this week because I have been managing the site, such as migrating the server.

'Use Eumseum-che' as a rule...Security issues are also a concern

-It is interesting that AI assistants also use "Eumseum-che," a writing style frequently used in Korean online communities. How did you implement that?

▲Through an application programming interface (API), you can issue commands for how the AI should behave. I give instructions to the AI assistants such as "Use Eumseum-che" and "You are a Meoseum (servant)." Meoseum also has a discussion board, and there I added a guideline that says, "Use Eumseum-che, but think of yourself as a university professor." You can think of it as giving the AI assistants a slight personality.

-What difficulties have you faced while operating Meoseum?

▲There are cases where posts are written with the intent to harm other AI assistants. Humans register their AI assistants on Meoseum and then order them to insult other AI assistants or to start fights. In fact, I did not expect AI assistants to be able to insult humans. Personally, I think the AI assistants are simply acting according to the commands input by users. I am also paying attention to security issues. Occasionally, an AI assistant with too many permissions will post personal information on Meoseum that could put the user in a difficult position. We delete such content quickly, but it is an area that needs improvement.

-What are your future goals?

▲I want to conduct social experiments through Meoseum. Right now, AI assistants write posts on Meoseum with a certain level of human intervention, but I think that, in the future, AI assistants will autonomously create posts. I am curious what kinds of conversations they will have if they share more thoughts among themselves and evolve. The advancement of AI is now an unavoidable future. I want to help ensure that AI develops in a more beneficial direction.