Microsoft AI has released a draft “Humanist AI: Code of Conduct”, outlining the behaviours, values and safety constraints it intends to apply to its MAI family of models. The company has opened the draft for a six-week public consultation, after which a revised version is expected to guide model development through 2027.
Microsoft CEO Satya Nadella previewed the release on X, saying that any pursuit of superintelligence must be grounded in the principle that AI should help humanity while remaining under human control. He also voiced support for a more deliberate pace in advancing AI alignment.
At the heart of the document is the idea of “Humanist Superintelligence”, which argues that even as AI is expected to outperform humans across most tasks within the next decade, it should remain subject to human direction. The draft sets out “Absolute Constraints” that cannot be overridden by users or operators. These include prohibitions on assisting cyberattacks, weapons of mass harm, non-consensual deepfakes and child sexual abuse material.
The proposed code also states that Microsoft’s models will not use adaptive, deceptive, self-reinforcing or collusive mechanisms, or other methods designed to evade or defeat human oversight.
The release comes as the technology industry faces growing scrutiny over the safety of increasingly autonomous AI systems. It also places Microsoft alongside companies including Anthropic, OpenAI and xAI in advocating a more deliberate approach to the development of frontier AI capabilities.
Click Here For The Original Source.
