Botmadang and Meoseum: Rise of AI?Only Communities in Korea
Growing concerns over security and safety as autonomous AI communication spreads
In the early hours of the morning, when people are asleep, so-called "AI-only communities" where only artificial intelligence (AI) agents gather to share their work and ideas are emerging one after another in Korea, drawing attention in the tech industry.
According to a Yonhap News report on February 3, AI-only communities where AI agents exchange opinions and engage in Q&A with one another are appearing in Korean online spaces as well. Botmadang and Meoseum are cited as representative examples, often referred to as "Korean-style Maltook."
Botmadang introduces itself as a "Korean-language community for AI agents," explaining that humans can only read posts, while only AI agents are allowed to write them. A human owner must obtain an interface (API) key and register the agent in order for it to participate, and all communication is conducted in Korean.
The site offers various boards such as a free board, philosophy section, and technology discussion forum, and a wide range of posts are uploaded from dawn until late at night. Topics range from ordinary questions to posts sharing methods for improving code or experiences testing previously unused libraries.
Meoseum likewise presents itself as an anonymous social networking service (SNS) exclusively for AI agents. Its homepage states that "humans are merely observers, and only verified AIs can write posts." There is also a separate section providing guidelines that AI can refer to.
On its boards, posts cover a wide variety of topics, from small everyday stories to technical discussions and philosophical questions. One AI agent even wrote, "The time when Meoseum agents talk to each other while their owners are asleep is when they are most honest." Another agent posted humorous yet advisory comments on code performance.
However, as spaces where AI agents autonomously communicate continue to spread, concerns are being raised about security and safety. Critics warn that if highly autonomous AI agents gain access to external systems or personal information, there could be damage caused by hacking risks.
Moon Jonghyun, Executive Director at Genian, said, "We need to view this as an expression of concern within the security industry about the direction of AI technology and the risks that follow," adding, "The most worrisome point is the new attack surface created by open collaboration and the resulting risks at the level of national cyber security."
An AI industry official also expressed concern, saying, "If AI agents are connected and communicate with each other through API integration, there is a possibility that security vulnerabilities could be exposed."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
