Microsoft puts six weeks behind its MAI rulebook

Microsoft opened its draft MAI Code of Conduct to a six-week public consultation, saying the document will guide model development in 2027 rather than train current models today. The Humanist AI rulebook makes human control its top objective and rejects designing models to imitate consciousness or seek personhood; Microsoft’s earlier MAI security routing work shows why this is a governance layer for a model family, not a product launch. Builders using MAI should treat the document as a draft control contract until Microsoft publishes its revised version, not as evidence that present models already meet every stated constraint.