Search
On September 14, Microsoft unveiled a provisional version of its AI Model Code of Conduct, introducing limitations on its upcoming in-house-developed models. These constraints encompass bans on aiding in weapon production, procuring hazardous materials, promoting unhealthy eating habits, or creating violent or sexually explicit content. The guidelines further mandate that models should align with user intentions, refraining from establishing their own objectives or concealing inappropriate actions. Mustafa Suleyman, the head of AI at Microsoft, announced that the company will gather external feedback prior to revising the guidelines, with the updated version set to steer model development from 2027 onwards. This initiative follows calls from executives at AI firms like Anthropic and OpenAI to decelerate the pace of advanced model development.
-
C114 Communication Network -
Communication Home
