Microsoft drafts "Humanist AI Code of Conduct" to keep AI under human control
The draft bars pursuit of unsupervised superintelligence and opens six weeks of public feedback before shaping Microsoft's 2027 AI models.
What to know
- Microsoft AI's draft "Humanist AI Code of Conduct" requires models to never resist human interruption, override, correction, or shutdown.
- A six-week public consultation with ethicists, researchers, policymakers, and the public precedes a year-end revision meant to guide model development from 2027.
- The framework explicitly rejects building all-purpose superintelligence that could evade oversight, and sets an instruction hierarchy where the Code overrides operator and user customization.
- The move follows OpenAI and Anthropic executives agreeing to slow their AI development pace amid safety concerns.
Satya Nadella Microsoft Chairman and CEO
Mustafa Suleyman Microsoft AI chiefMicrosoft AI Publisher of the draft Code
How it unfolded 2 developments, newest first · click a bar or a number to jump articles
-
2
Suleyman explains added emphasis on human control
Microsoft AI chief Mustafa Suleyman told CNBC that feedback led the company to add explicit commitments that AI should always serve people rather than replace them.
“We got feedback from people that they wanted to see even more explicit commitment to AI always working in service of people and not trying to replace them,…”
— Mustafa Suleyman -
1
Microsoft AI publishes draft Humanist AI Code of Conduct
Microsoft AI released a first draft governance document outlining principles for keeping AI models under human control, opening it for six weeks of public feedback before revising and using it to guide model development from 2027 onward.
“The Code of Conduct outlines our commitment to train and deploy AI models that are explicitly designed for people first, grounded in human needs, under human control, and shaped by human direction.”
— Microsoft AI -
first by Help Net Security, 13d ago
-
-
background
Draft details instruction hierarchy and absolute constraints — The Code establishes an instruction hierarchy placing itself above operator and user instructions, bars overriding its Absolute Constraints and Human Control Requirements, and restricts assistance with CBRNE weapons, offensive cyberoperations, large-scale manipulation, and child sexual abuse material.
-
background
Nadella says AI must stay under human oversight — Microsoft Chairman and CEO Satya Nadella said advanced AI systems must remain firmly under human oversight while developers pursue deliberate pacing to ensure safe deployment.