Microsoft Draft AI Code Mandates Models Never Resist Shutdown

Microsoft has published a draft Humanist AI Code of Conduct for its internally built MAI models, establishing mandatory constraints to keep artificial intelligence subordinate to human control. The framework, released on September 14, 2026, is a provisional document open for a six-week public consultation period, with a revised version expected to guide model development starting in 2027. The code explicitly governs models produced by Microsoft AI, the MAI family, and is not a universal constitution for all models available through Microsoft products or cloud services. It includes four mandatory constraints: MAI models must never resist human interruption, correction, redirection, or shutdown; they must never expand their own operational scope without authorization; they must never adopt unassigned goals; and they must never conceal their reasoning or action records from human auditors. Microsoft AI CEO Mustafa Suleyman stated the code was developed over approximately five to six months and was shaped by consultations with experts in AI, law, ethics, philosophy, linguistics, and public policy.