Microsoft has published a 37-page document titled Humanist AI Code of Conduct on September 14, 2026. The framework establishes that AI models must remain under human control, avoid simulating consciousness, and refrain from seeking legal personhood.
The code defines any loss of human control as a system failure. It introduces the concept of Humanist Superintelligence, which prioritizes safety by intentionally limiting model autonomy. This approach contrasts with general superintelligence models by treating constraints as design features.
The development of the document spanned five to six months with input from various internal teams. Microsoft has initiated a six-week public consultation period to gather external feedback before finalizing the policy.
This initiative builds upon the company's existing Responsible AI principles, which include fairness, reliability, safety, privacy, security, inclusiveness, transparency, and accountability. The new code provides specific protocols to implement these pillars.
The concept was first proposed by Microsoft AI CEO Mustafa Suleyman in November 2025. Additionally, CEO Satya Nadella emphasized on September 13, 2026, that AI must serve humanity, following a transparency report released by the company on September 1.
