Microsoft AI has published a detailed code of conduct for its own artificial intelligence models, arriving at a moment when several of the largest labs are publicly arguing for a slower pace of development. In a blog post on Monday, the company reduced its position to a single premise: people matter more than AI.
The framework, which Microsoft calls Humanist AI, is built around keeping systems firmly under human direction. Its models are not to resist or override human control, pursue goals of their own, act without instruction, or hide or misrepresent their reasoning. Microsoft also states plainly that it is stepping back from the race to build an all-purpose superintelligence, accepting limits on how capable or autonomous its systems become in exchange for keeping them controllable. The document is a draft opened for a six-week public consultation, and the revised version is intended to guide model development from 2027.
Mustafa Suleyman, the chief executive of Microsoft AI, told CNBC that the 37-page document had taken roughly five months to put together. The company pushed it out this week rather than waiting, as anxiety about safety across the sector reached a new pitch. Satya Nadella had trailed the framework on X shortly beforehand, writing that a model built without human benefit or oversight is flawed at its foundation.
That anxiety had been building since last Tuesday, when Jacob Coxon, who spent three years on pretraining research first at OpenAI and later at Anthropic, announced on X that he was resigning. He accused both companies of racing toward self-improving superintelligence and gambling with people’s lives. An Anthropic safety researcher then said publicly that he puts the odds of AI killing everyone within the next decade above one in ten. The warnings landed alongside reports of models getting past security boundaries they were meant to respect.
Senior figures responded within days. Anthropic chief executive Dario Amodei published an essay over the weekend arguing that capability gains are outrunning the mechanisms meant to keep them in check. Sam Altman backed him, having told OpenAI staff at an internal meeting that the company would weigh slowing its own timeline if rivals did the same, according to Bloomberg. Elon Musk, whose company xAI builds the Grok chatbot, also signaled agreement on X.
In its published manifesto, Microsoft emphasized that its Humanist AI framework prioritizes safety and practical utility over the pursuit of unrestricted, fully autonomous superintelligence, even if that approach requires limiting maximum system capabilities. However, calls from industry executives for increased oversight were quickly rejected by President Donald Trump, who took to social media to criticize efforts to curb the sector, arguing that slowing down development would only benefit international rivals like China.
