MS Unveils Draft Human-First AI Code of Conduct

Microsoft (MS) has unveiled a draft code of conduct that includes the principle that it will not create artificial intelligence that humans cannot control. The announcement is interpreted as a message that, amid the continuing race to develop AI, it will place greater emphasis on safety and controllability. On the 14th (local time), Microsoft AI, MS’s AI division, released a 37page draft of the “Humanist AI Code of Conduct” and said it would place at the core the principle that “people come before AI.” The company also explained that it can restrict AI performance and autonomy if necessary. The draft also includes a commitment not to participate in the race to develop generalpurpose superintelligence that can break through safeguards. MS said AI must remain under human direction and supervision, and that communication between AIs in ways humans cannot understand will not be permitted. This policy also aligns with the safety frameworks currently operated by other AI companies such as Anthropic and Google DeepMind. However, since the 기준 for distinguishing “uncontrollable AI” is still unclear, attention is on whether discussions will continue to unify safety rules among companies into a common standard.