Microsoft publishes draft Humanist AI Code of Conduct for MAI models

Microsoft AI released a draft Humanist AI Code of Conduct for public consultation, intended to become the primary governing document for its MAI models from 2027. The code prioritises human control, prohibits self-directed goals and opaque agent communication, and sets absolute constraints on assistance with weapons, deception and shutdown resistance. The six-week consultation reflects industry movement toward explicit, publicly stated behavioural limits on frontier systems.