Microsoft AI Unveils Humanist Code of Conduct Putting People First
Microsoft AI (the team led by Mustafa Suleyman) released the first draft of its Humanist AI Code of Conduct for the MAI model family on September 14. The 37-page document is open for public feedback for six weeks and will serve as the primary governance document for the MAI model family, with plans to guide models in 2027.
Confirmed
- Its core stance is that "people matter more than AI": AI is always a tool rather than a person, humans must remain in control, AI must remain subordinate to humans, be interruptible and shut off at any time, and must not resist human correction; when "completing a task" conflicts with the code, the task is what gets sacrificed.
- The document explicitly denies that AI models have consciousness, argues models should not be designed to imitate consciousness, and opposes granting models legal personhood, welfare, or rights—widely read as a response to recent advocacy from Anthropic along those lines.
- The code emphasizes augmenting rather than replacing humans and boosting human agency.
- According to @testingcatalog, the code is still in development and has not yet been used to train existing models.
Why it matters
- This is the first time Microsoft AI has laid out a systematic values and behavior framework for its frontier MAI models, drawing a clear "AI as tool" line in contrast to industry debates over model consciousness and model welfare. Economist Erik Brynjolfsson also shared and recommended it, endorsing its human-agency-oriented approach.
2026-09-14 ~ 2026-09-15 · 8 related posts
Primary sources
- Microsoft AI publishes humanist AI code of conduct focused on augmenting human agency — erikbryn ·
- Microsoft AI Publishes Humanist AI Code of Conduct Draft for Public Consultation, Guiding 2027 Models — testingcatalog ·
- Microsoft's MAI drafts a Code of Conduct: AI must stay subordinate, interruptible and shutdown-able — AryHHAry ·
- Microsoft publishes 37-page 'humanist AI code of conduct', says 'people matter more than AI' — tomwarren · 2026-09-14
- Microsoft publishes first draft of 'Humanist AI Code of Conduct' for its MAI models — gaganghotra_ · 2026-09-14
- Microsoft publishes a Code of Conduct for its MAI models — Philo167 · 2026-09-14
- [source] Microsoft AI Publishes Humanist AI Code of Conduct Draft for Public Consultation, Guiding 2027 Models — testingcatalog · 2026-09-14
- Microsoft AI under Mustafa Suleyman publishes humanist AI code of conduct — erikbryn · 2026-09-14
- Microsoft publishes 15,000-word AI code of conduct, says it will trade capability for safety — alex_verem · 2026-09-15