Microsoft has published a 37-page “humanist AI code of conduct” as the safety of advanced AI systems comes under debate. The document states that humans are more important than AI, that models must remain under human oversight under all circumstances, and that they are not conscious. Microsoft also rejects granting models legal personhood or rights, as well as safeguarding their welfare. This approach differs from Anthropic’s research into AI consciousness and model welfare. While Anthropic CEO Dario Amodei says advanced models could be conscious, Microsoft AI CEO Mustafa Suleyman described these assessments as “really very dangerous” in June. As it competes with Google, Anthropic and OpenAI, Microsoft advocates preserving human oversight, ensuring that tasks that would violate safety boundaries fail, and not joining a superintelligence race that could bypass safety measures.
Why it matters
The document frames the debate over the status of AI systems not only in terms of technical performance but also through the relationship of authority they will establish with humans. Microsoft’s approach differs from Anthropic’s position on whether models should be considered conscious or capable of welfare, showing that the industry lacks a common ethical foundation. The principles of preserving human oversight and ensuring that tasks violating safety boundaries fail demonstrate, for the parties developing and using these systems, that responsibility remains with humans rather than being transferred to the model. The open question is how the differing views of consciousness, rights and safety held by various companies for advanced models will translate into rules, and whether they will establish a common standard.
Background
Microsoft is not a new name in the FikirPilot archive: over the past 90 days, we have published 11 news stories mentioning the name; the latest is dated September 14, 2026.