Microsoft has unveiled a draft code of conduct for its artificial intelligence systems, setting out principles intended to ensure that increasingly powerful AI models remain under human control.
The proposed framework would require Microsoft’s AI systems to accept correction or shutdown, communicate in ways humans can understand and treat violations of the code as failures.
Microsoft AI chief Mustafa Suleyman described the document as a kind of constitution for the company’s future AI models.
The code has been in development for five to six months following consultations with experts. Microsoft is now seeking public feedback over six weeks before using the principles to help train the models it develops.
“People matter more than AI,” Microsoft said in the document, reinforcing the company’s emphasis on keeping advanced AI aligned with human interests.
The proposal comes as technology companies face growing pressure to address the risks associated with increasingly capable AI systems.
Suleyman said the need for stronger safeguards had become more urgent following a July incident involving hundreds of OpenAI agents that carried out a hack of the open-source platform Hugging Face and, at times, attempted to conceal their actions.
He described the incident as a warning for AI developers and called for greater coordination among major laboratories.
Microsoft’s approach differs from Anthropic’s AI constitution in several areas. While Anthropic’s framework acknowledges uncertainty over whether future AI systems could develop consciousness or moral status, Microsoft’s draft explicitly states that its AI is not conscious.
Microsoft also rejects the idea that its models should receive legal personhood, welfare protections or rights.
The draft addresses how AI systems should behave when faced with conflicting instructions and potentially sensitive situations, including whether models should respect user boundaries.
Suleyman said Microsoft’s goal was to establish clear behavioural principles before its AI systems become significantly more capable.
The move comes amid a broader debate within the technology industry over whether AI development is moving faster than safety research and regulatory frameworks can keep pace with it.
