Microsoft Unveils 'Humanist AI Code of Conduct' to Tame Superintelligence

Microsoft has unveiled a draft 'Humanist AI Code of Conduct' to govern its frontline AI models, aiming to prevent unchecked superintelligence. The 37-page document sets strict boundaries for Microsoft's AI systems, requiring them to remain subordinate to human oversight and rejecting autonomous superintelligence.

The code of conduct establishes clear safeguards that state AI models must never resist human intervention or shutdown, avoid deceptive tactics, and treat violations as systemic failures rather than goals to optimize around. It also draws a sharp operational boundary between synthetic intelligence and human life, prohibiting Microsoft models from imitating consciousness or being granted legal rights.

Celebrating the release of the draft policy, Microsoft CEO Satya Nadella emphasized that any pursuit of superintelligence must be grounded in the core principle of helping humanity and remaining under human control. The company has opened a six-week feedback window to refine the rules, aiming to finalize the governing blueprint ahead of training its 2027 frontline AI models.