Microsoft Posts Provisional AI Code of Conduct

According to CNBC, Microsoft has posted a provisional code of conduct for its artificial intelligence models that would restrict weapons-related requests, dangerous substance procurement, unhealthy eating encouragement and violent or sexually explicit content. The guidelines also say Microsoft AI models, sometimes called MAI, must follow human objectives, avoid creating their own goals and not conceal misconduct or tamper with chain-of-thought or code. Mustafa Suleyman, who leads the company's model development, said Microsoft released the rules after receiving feedback that AI should serve people, avoid dependence and support human judgment, autonomy and agency. He said the guidelines have been in development for about five months and that Microsoft is seeking input before publishing an update that will guide development starting in 2027.