Microsoft Publishes its MAI Code of Conduct — Humans Must Remain In Control!
The key takeaway here is that MAI models need to be controllable through human readable explanations, interruptible, shut downable, non-self-improving, and not granted rights or “inner lives”.
The Microsoft Code of Conduct for their MAI models emphasizes human control, using explanations (in human readable format) as the primary means of doing so. Models must also be interruptible/shutdownable and non-self improving. Models cannot use “neurolanguage” to evade oversight. Finally, they do not have rights nor are they granted inner lives.
If the explanations are opaque, then the model is uncontrollable. Note that this was published shortly after GPT-6 Astra, which provides reasoning explanations that are harder to follow compared to its predecessors. For agentic models, transparency of thought processes is essential rather than optional.
Source: the-decoder.com
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →▌ Related guides
Author
Evgenii Arsentev
PhD · Chief Executive Officer, digital health
Articles · Latest articles