Microsoft Publishes Code for MAI Models Amid Calls for AI Pacing

Microsoft's CEO Satya Nadella has announced that the company will publish its 'Code of Conduct' for first-party MAI models on September 14, 2026, for public consultation. This move comes in response to a recent essay by Anthropic CEO Dario Amodei, who argued that the rate of AI capability improvement must be slowed down to prevent risks.

Nadella stated that any pursuit of superintelligence must be grounded in the principle that the AI built should help humanity and be under human control. He welcomed the research and focus on making alignment the design goal, along with ideas like embedded evaluators.

The MAI models, introduced by Microsoft AI in June 2026, include image, voice, transcription, coding, and reasoning capabilities. Nadella emphasized that these models should be available for broad access and choice at every layer of the AI stack, with enterprise control over learning loops and models.