Microsoft drafts feel-good AI model guidelines and wants your input
Redmond outlines 'aspirational' goals for model behavior, but gives itself a pass if it gets things wrong
Microsoft has introduced draft principles for governing the development of future AI models, inviting public feedback to refine its Humanist AI Code of Conduct. The company's primary objective is to ensure humans maintain control over AI to help people lead healthier, happier, and more productive lives. The guidelines cover technical controls, operational and monitoring systems, and organizational culture, focusing on Microsoft AI (MAI) models to promote health, happiness, and productivity.
Microsoft defines superintelligence as AI systems more intelligent than all humans combined, though the term's definition leaves intelligence undefined. The company's current models include MAI-Transcribe-2, MAI-Thinking-1, MAI-Code-1.1-Flash, MAI-Image-2.6, MAI-Voice-2, and MAI-Cyber-1-Flash. While not currently among the most intelligent models, future Microsoft models could potentially be capable enough to generate concern over AI behavior.
Microsoft has timed its human primacy manifesto well, given the unwillingness of other AI developers to practice self-restraint. However, the Humanist AI Code of Conduct has no teeth and is designed to cover Microsoft employees and contractors, not customers. The document should be read as a north star for development and training, not a guarantee of present-day performance. The company will accept public input until October 25, 2026.
Written by urgent.news from The Register Science's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
