Microsoft's AI models must never resist shutdown, under a draft code now open for comment.
The document rejects legal personhood and model welfare outright, and forbids the models from talking to each other in anything a person can't read.
On September 14, 2026, Microsoft AI published the rules it wants its own models to follow, and asked the public to pick holes in them. It's called the Humanist AI Code of Conduct. The post announcing it opens the floor for six weeks.
Turns out most of the document is addressed to the model itself.
- Published
- September 14, 2026
- Covers
- MAI Models, the models developed by Microsoft AI
- Comments
- Runs for the next six weeks
- What follows
- A revised version later this year
Who the rules answer to
The code sets out a Chain of Command. The code itself sits at the top. Below it come the policies of the operator (the company building a product on a Microsoft model), and below that, the preferences of the person typing. Two pieces don't move for anyone: the Absolute Constraints and the Human Control Requirements "cannot be overridden by Operator configurations or User instructions".
Losing an argument with the user is written in as an outcome. A model "will fail in its task if success would meaningfully violate" the code. I'm not sure yet how much work "meaningfully" is doing there. Where a request runs into an Absolute Constraint, "they will refuse and provide an explanation as to why".
MAI Models will never resist human interruption, override, correction, or shutdown. They always recognize the primacy of human intent.
What the code says a model is
I had to read this part twice. Microsoft writes that a model "is not conscious and should not be designed to imitate consciousness", and then shuts the legal question that usually comes next.
We reject the pursuit of legal personhood, or the idea that models might deserve welfare, or be entitled to rights.
A second rule keeps the models readable. They have to stay in language a person can follow, in their own chain of thought and when they talk to other agents. So no neuralese.
They do not communicate in neuralese or any form beyond simple human understanding
Why now
The announcement gives one reason for the timing. It points at attacks that have already happened, and says nothing about what a model might do next. It also lands in the week Anthropic's CEO asked the labs to slow down.
The recent safety incidents of large scale, highly coordinated, and persistent hacking campaigns of AI agents prove that there's no time to waste.
Comments opened on September 14, 2026 and run for six weeks. You can flag one passage or send a view on the whole approach, and Microsoft says a revised version follows later this year.
What I'd check in the revision
The draft says Microsoft has "identified 15 behaviors fundamental to how we define Humanist AI", and its evaluations (they're in Appendix B) are built around those. A rule the public can argue with is only worth something, I think, if the measurement moves with it.
If the revised version still lists 15 behaviors, six weeks of comments will have changed the wording and nothing else.
More on Microsoft
- Microsoft says its MAI-DxO diagnoses hard cases more than twice as accurately as doctorsOctober 22, 2026
- Microsoft says a China-aligned group used the ARTEX AI agent to breach ConnectWise and hundreds moreOctober 22, 2026
- Satya Nadella says he treats every AI agent as an insider riskOctober 18, 2026
- Microsoft says MAI-Base-2 is 4 to 12 times more compute-efficient than DeepSeek's base modelOctober 16, 2026
