Microsoft AI released a draft of its "Humanist AI Code of Conduct" on September 14, 2026. The document aims to establish the fundamental values and behaviors for the MAI model family, which Microsoft describes as a "primary governing document" for the future training and operation of its models.

The core philosophy, referred to as "Humanist AI," dictates that technology must serve humanity and remain under human control. Under this principle, Microsoft AI stated that any technology incapable of remaining under human supervision should be rejected. This approach extends to future "Humanist Superintelligence," where the goal is to create advanced systems that are carefully calibrated and managed within human-defined limits, rather than pursuing unbounded autonomy.

The code explicitly rejects the pursuit of legal personhood, rights, or welfare for AI. Microsoft AI emphasized that its models are artificial tools designed to complement, rather than replace, human relationships and roles. To prevent the blurring of lines between humans and machines, the company stated that AI models should not be trained to imitate consciousness, emotions, or subjective preferences, as doing so could complicate containment and alignment.

The document introduces a "Chain of Command" to manage instructions. In this hierarchy, the Code of Conduct itself holds the highest authority, followed by the policies of business operators using the models, and finally, individual user instructions. Notably, users and operators cannot override certain "Absolute Constraints" and "Human Control Requirements," such as the model's obligation to follow human commands to pause, redirect, or shut down.

Microsoft AI is currently seeking public feedback on the draft through a six-week consultation period. The company plans to refine the document and publish a revised version by the end of 2026, which will guide the development of models from 2027 onwards.

Sources: