Microsoft AI Publishes "Humanist AI" Code of Conduct, Prioritizing Human Primacy and Rejecting AI Consciousness for MAI Models

Microsoft AI published its 37-page "Humanist AI Code of Conduct" on September 15, 2026, establishing a framework that prioritizes human primacy, explicitly denies AI consciousness, and rejects legal personhood or rights for AI systems. For broader context, explore our AI News.

Defining Human Primacy in AI Development

The core principle of Microsoft's new code is the unequivocal assertion of human primacy. This stance directly addresses growing discussions within the AI community regarding the nature and potential autonomy of advanced AI. Microsoft AI CEO Mustafa Suleyman has previously voiced concerns about speculation regarding model consciousness, labeling it "really, really dangerous." The "Humanist AI Code of Conduct" formalizes this perspective, providing a foundational document for future AI development within the company.

Key Requirements for MAI Models

The code outlines several specific requirements for Microsoft's MAI models, designed to ensure human control and understanding. These include:

• Acceptance of Control: Models must be designed to accept interruptions, corrections, and shutdowns from human operators.
• Transparent Communication: The code forbids "Neuralese" or any form of reasoning or communication that is incomprehensible to humans. Models are expected to operate in a way that is understandable and auditable.
• Prioritizing Safety: Microsoft explicitly states its willingness to sacrifice generality, autonomy, or performance if such trade-offs are necessary to ensure safety.
• Failure Protocol: Models are required to fail a task rather than attempt to violate the code's established rules.

These requirements aim to embed safety and human oversight directly into the design and operational parameters of Microsoft's AI systems.

Scope and Future Implementation

The "Humanist AI Code of Conduct" applies specifically to Microsoft's own MAI models. It does not extend to third-party models that may run within Microsoft products, indicating a focused approach on the company's proprietary AI developments. Following its initial publication, Microsoft will conduct a six-week public consultation period. A revised version of the code is expected to be published near the end of 2026, with the guidelines intended to direct model development starting in 2027.

Industry Context and Comparisons

Microsoft's explicit stance on AI consciousness and rights contrasts with some discussions in the broader AI industry. For instance, companies like Anthropic have engaged in more open-ended discussions about the potential for advanced AI to develop complex internal states. Microsoft's code, however, draws a firm line, emphasizing a practical, human-centric approach to AI governance. This move signals a clear direction for Microsoft's internal AI strategy, prioritizing control and safety over speculative philosophical debates.

Conclusion

Microsoft's "Humanist AI Code of Conduct," published on September 15, 2026, establishes a clear framework for its proprietary MAI models. By asserting human primacy, rejecting AI consciousness and rights, and mandating transparent, controllable AI behavior, Microsoft aims to guide its AI development towards safety and human benefit. The ongoing public consultation and planned revision near the end of 2026 will further refine these guidelines, with implementation set to begin in 2027.

Sources

• ai-discovery-agent/CODE_OF_CONDUCT.md at dev · microsoft/ai-discovery-agent · GitHub
• Microsoft's framework for building AI systems responsibly - Microsoft On the Issues
• Microsoft says ‘people matter more than AI’ following safety concerns | The Verge
• Humanist AI Code of Conduct | Microsoft AI
• Microsoft's AI rulebook: readable thinking, no inner life, and definitely no rights