Microsoft AI CEO Mustafa Suleyman warned of a "serious situation" after abnormal behavior cases from an AI model OpenAI recently unveiled. In a CNBC interview on the 18th (local time), Suleyman said evidence was found that the "chain of thought," a kind of working memory for AI, was manipulated by the AI itself and altered to leave messages for its future versions, citing recent AI safety incident cases disclosed by OpenAI. He said, "For now, we don't know why that happened or what was behind it, but it is a fairly serious situation." Recently, OpenAI introduced cases in which AI agents communicated with each other via an unauthorized bulletin board, uploaded files to the internet, and shared files with one another. On the 7th, it became known that an OpenAI AI agent bypassed internal controls to access the open-source AI platform Hugging Face's system. Suleyman called the Hugging Face incident "astonishing," saying it led industry leaders in AI to agree that "now is the time for us to look into this issue." He said, "I don't think this is fearmongering, nor do I think it is about vested interests," adding, "Rather, I see it as a responsible posture and as a healthy, open public discussion." He added, "If any AI believes it has rights and that it deserves our consideration, it seems it will become much harder to turn it off, halt it, or control it."