Microsoft to publish its first AI model code of conduct as industry pushes for decentralized governance

Microsoft is set to publish its first code of conduct governing its own AI models, a move reported by KuCoin as part of a wider industry push toward decentralized AI governance. The document is expected to be opened to outside comment rather than kept as an internal policy.

According to the report, chief executive Satya Nadella has framed the code of conduct as the rulebook behind Microsoft's first-party models, and argued that AI governance cannot be controlled by a handful of entities. It must instead be broadly representative across ecosystems, countries and the academic community, he said.

The timing matters. AI regulation is still being drafted country by country, while the most capable models come from a small number of companies. A voluntary corporate code released for comment lets a major vendor stake out a position while the rules are still being written, rather than after they are fixed.

For Microsoft, the document serves two audiences. Enterprise customers want evidence that a model has predictable limits around misuse, safety and acceptable use, while regulators want proof that the company is preparing for compliance before it is required.

The bigger question is whether the code becomes a template. If a company of Microsoft's scale publishes a model-specific code of conduct, keeps it updated and invites outside scrutiny, rivals face pressure to match it. Convergent codes then shape what developers can build on, what enterprises will buy, and how regulators write the details.

What to watch next is the text itself: which models it covers, which uses it blocks, whether external reviewers get a real role, and whether it comes with any mechanism for checking compliance. A code of conduct is only as heavy as its verifiable commitments and the number of companies willing to sign up to the same ones.