North Korea appears to have developed its first AI chatbot, according to an analysis by the monitoring group 38 North of footage broadcast on North Korean state television, reported by Reuters early Thursday. The system, called “AI Service System 1.0, Fountain of Wisdom,” answers science and technology questions in Korean and English — and carries an on-screen warning telling users not to ask socio-political questions. Why it matters: it is a rare, concrete look at how one of the world’s most closed states is adopting AI — on a foreign-built foundation, behind a political filter.
November 19, 2026 — Tested Intelligence News
What happened
38 North, a think tank that monitors North Korea, identified the chatbot in footage from the National Exhibition of IT Successes in Pyongyang, held from October 19 to 30. The system featured in a report broadcast on Ryongnamsan TV on November 17, and Reuters reported the analysis on November 19.
According to 38 North, the chatbot was built by Kim Il Sung University’s Artificial Intelligence Technology Research Institute together with Phurunhanul Trading Co, a Pyongyang software developer. It can answer questions on science and technology in Korean or English and provide translations.
The details that stand out
Two details in the footage say more than the demo itself. First, an on-screen warning instructs users not to ask socio-political questions, and the footage appears to show a classifier screening queries before they reach the AI engine — a guardrail layer, in other words, tuned to politics rather than safety. Second, the hardware running the system was not disclosed.
38 North’s assessment is that the model behind the interface is more likely an open-weight system from a US or Chinese laboratory — such as Google’s Gemma or Alibaba’s Qwen — than one developed entirely in North Korea. The report also noted that the country’s lack of large-scale data centres could hamper deployment of the most advanced models.
Why it matters
For anyone tracking how AI spreads, this is the pattern to watch: open-weight models travel everywhere, including into sanctioned, largely offline states, and the local work is in the wrapping — the interface, the language support, and above all the filters. North Korea’s version answers science questions and translates, while a screening layer keeps politics out. That is a very different control model from the safety guardrails Western labs argue about, and it underlines why export debates increasingly focus on open weights: once a capable model is downloadable, who runs it and what they bolt onto it is largely out of the original lab’s hands.
Keep the scale in perspective. This is a chatbot shown at a domestic IT exhibition, identified from state TV footage by outside analysts — not a deployed national service, and none of its capabilities have been independently tested. But as a signal of where even the most isolated regimes are heading, it is a clear one.
Frequently asked questions
What is North Korea’s AI chatbot called?
“AI Service System 1.0, Fountain of Wisdom,” according to 38 North’s analysis of state television footage from an IT exhibition in Pyongyang, reported by Reuters on November 19, 2026.
Who built it?
Kim Il Sung University’s Artificial Intelligence Technology Research Institute and Phurunhanul Trading Co, a Pyongyang software developer, according to 38 North. The underlying model is more likely an open-weight system from a US or Chinese lab, such as Google’s Gemma or Alibaba’s Qwen, than a fully homegrown model.
What can the chatbot do?
Based on the footage, it answers science and technology questions in Korean or English and provides translations. An on-screen warning tells users not to ask socio-political questions, and queries appear to be screened by a classifier before reaching the AI engine.
Sources: Reuters, 38 North

