Microsoft has published a draft AI Code of Conduct and opened a six-week public consultation on how it should develop advanced AI. The central idea, as described by AI Magazine, is “Humanist AI”: systems should remain subordinate to people, aligned with human purposes and subject to meaningful oversight.
Watch Desk analysis
What happened
The draft is framed as a starting point for Microsoft's approach to superintelligence, which the company describes as a technology with consequences far beyond any single firm. Microsoft says it wants public feedback on the principles behind that approach, with the consultation running for six weeks.
The supplied account does not provide the full draft text, detailed enforcement mechanisms or evidence that the proposed principles are already applied consistently across Microsoft's products. Those are important gaps, because “human control” can mean anything from a real operational brake to a phrase polished for a keynote.
Why it matters
This gives the AI safety debate a concrete corporate document to inspect. Microsoft is not merely saying that advanced systems should be helpful. It is stating that they should not outrank human authority or become people-shaped actors in their own right.
For readers, the useful question is what the code turns into when systems can act, spend, decide or operate at scale. Principles matter, but the plumbing decides whether they survive contact with reality.
Our read
Worth reading as a proposal, not accepting as proof that Microsoft's safeguards are complete. The consultation is the opportunity to test the language against difficult cases: who has the final say, how quickly can a system be stopped and what happens when commercial incentives pull the other way?
What to watch
- Whether Microsoft publishes the full draft and responses to consultation feedback.
- What concrete tests, approval gates or shutdown controls sit behind “human oversight”.
- Whether the principles apply across Microsoft's own products and its wider AI partnerships.
- How the code handles systems that can take consequential actions without waiting for a person.
Discussion spark: What specific control would convince you that an AI system is genuinely subordinate to human oversight rather than merely described that way?
Sources and evidence
- AI companies must relinquish some of the power at their disposal – Financial Times (14 September 2026, 14:12 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.