North Korean cybercriminals have turned to artificial intelligence (AI) to advance their spear-phishing efforts targeting DPRK-focused experts and organizations, Microsoft and OpenAI announced Wednesday, a move that one expert called “frightening.”

Microsoft reported that it observed North Korean threat group Kimsuky, also known as “Emerald Sleet,” using large language models (LLMs) to research potential targets with expertise in DPRK defense and nuclear issues and generate content for phishing campaigns.