Skip to content
NewsRegulation

Microsoft opens MAI Code of Conduct to public comment as Nadella backs deliberate AI pacing

· by Pondero Newsdesk

The short version

Microsoft released its first behavioral specification for first-party MAI models on September 14 for public consultation, with CEO Satya Nadella calling for independent evaluators and deliberate pacing on superintelligence.

Microsoft opens MAI Code of Conduct to public comment as Nadella backs deliberate AI pacing

No major hyperscaler had published a public-comment behavioral specification for its own first-party AI models before September 14, when Microsoft opened the document it calls the MAI Code of Conduct to external feedback. CEO Satya Nadella backed deliberate pacing and independent evaluators for superintelligence in the same announcement, joining a narrowing safety consensus among lab leaders.

What

Nadella announced the document via X on September 13, writing that "Any pursuit of superintelligence has to be grounded in the core principle that if the AI we build is not helping humanity and under human control, it's not worth pursuing," per Anadolu Agency reporting via ANNews. He welcomed "deliberate pacing needed to get alignment right" and proposed "embedded evaluators" to monitor AI systems inside laboratories.

The Code of Conduct governs Microsoft's first-party MAI model family, which spans five models released in June 2026 per Unite.AI: MAI-Thinking-1 for reasoning, MAI-Code-1-Flash for agentic coding, MAI-Image-2.5, MAI-Transcribe-1.5 supporting 43 languages, and MAI-Voice-2 supporting 15 languages.

Nadella described three pillars for the document: broad access and choice at every layer of the AI stack, enterprise control over continuous learning architectures and model weights without lock-in to a single developer, and the Code of Conduct itself. He also stated that governance "cannot be controlled by a handful of entities, but must have broad representation across the ecosystem, countries, and fields, including academia," per the same Anadolu Agency report.

Why it matters

For enterprise teams running MAI-based Copilot workflows, the Code of Conduct is the first formal behavioral specification they can read, challenge, and comment on before it closes. That comment window is the only formal channel to flag gaps now, before the document becomes the governing standard for those models.

Nadella's enterprise-control pillar has a direct operational implication: organizations should be able to run continuous learning systems and retain control of their own model weights rather than depending on Microsoft. That is a vendor-stated principle, not a regulatory requirement, but it echoes language procurement teams are increasingly inserting into AI vendor contracts.

The move also puts Microsoft on record with a named governance artifact before any regulator demands one. AWS and Oracle have not published equivalent open-comment documents for their first-party model families, as of September 14, 2026.

Context and reactions

The announcement landed inside a broader industry alignment on AI safety. Anthropic CEO Dario Amodei had proposed embedding independent reviewers inside laboratories, and OpenAI CEO Sam Altman indicated his firm would adopt the practice shortly after. Nadella aligned with both positions without adopting the specific "slowdown" framing, describing the moment as a call to translate safety concepts into "concrete operational mechanisms," per the Anadolu Agency report.

Microsoft published the MAI model family in June 2026 without an accompanying behavioral specification. The Code of Conduct fills that gap five months later.

What to watch next

Whether the public consultation produces substantive revisions will be the first signal of whether the process functions as a governance mechanism or a legitimacy exercise. Key milestones to follow: the close of the comment window, any published revision notes from Microsoft, and whether AWS or Google Cloud follow with equivalent open-comment documents for their own first-party model families.

Sources