Microsoft publishes 37-page 'humanist' code of conduct after AI doom debate: 'This is urgent'

Microsoft AI CEO Stephen Brashear has unveiled a 37-page draft code of conduct for its AI models, a move seen as urgent amid growing debates over AI's potential risks to humanity. The code, which Suleyman described as a "first draft for public consultation," prioritizes people over artificial intelligence, emphasizing that "People matter more than AI."

The document is structured into five key parts: safety constraints, guidelines for uncertainty, and model defaults. At the top of the hierarchy are the code of conduct, which cannot be violated, followed by operator policies and user preferences. Microsoft's models must adhere to strict safety protocols, including respect for intellectual property, internet access, data handling, and environmental boundaries.

The code also mandates that models explain their limitations to users, particularly in uncertain situations, and should interpret ambiguity in line with the best interpretation of the code.

Microsoft has intentionally placed barriers to prevent its AI from surpassing human control, stating, "AI should not exceed human control." The company argues that while building AI that is useful and safe may limit ultimate generality, autonomy, or capability, it remains committed to creating technology that benefits humanity.