The 'Humanist AI Code of Conduct' sets strict rules for
Microsoft's frontier models, banning assistance with weapons, violence, or
self-directed goals while requiring transparency and obedience to shutdown
commands.
CEO Satya Nadella stressed that superintelligence pursuits must stay under human control to benefit humanity.
The draft opens for six weeks of public input before guiding development from 2027 onward, amid industry calls for cautious pacing on advanced systems.