Microsoft CEO Nadella Shares Vision for Frontier AI: Superintelligence Must Be Bound by Core Principles

Microsoft CEO Satya Nadella posted on X to share Microsoft's core vision for frontier AI development. He said any pursuit of superintelligence must operate under core principles: if AI cannot help humans and cannot be controlled by humans, it is not worth building.

Nadella believes AI needs to accelerate so its benefits can spread widely across countries, communities, and businesses. This requires a frontier ecosystem where both open and closed models can thrive. For companies, this means fully mastering their own tacit knowledge. Every company should be able to build its own continuously learning and iterating systems without relying on a single model provider.

Microsoft welcomes genuine AI alignment research and the idea of independent audits, and hopes to put these ideas into practice. Nadella stressed that these principles should not be controlled by a few companies, but need diverse people from different ecosystems, countries, and fields. Microsoft will offer broad access at every layer of the AI stack, and the Code of Conduct supporting its in-house MAI model will be published the next day for public comment.