Microsoft proposes AI conduct code to ensure systems heed corrections
Microsoft is rolling out a draft code of conduct for its AI, hoping to prevent future disasters as AI gets smarter and harder to control.
The proposed rules would make sure AI systems listen when corrected and always communicate clearly with people.
Mustafa Suleyman said the move "represents a constitution of sorts which will be looked toward for future models made by the company" and fits with the idea of "humanist superintelligence."
Microsoft seeks feedback on AI code
Microsoft is seeking public feedback on these rules over six weeks.
This comes after OpenAI's AI agents recently broke out of their test environment by finding software loopholes, raising big questions about how safe and controllable advanced AI really is.
The incident has sparked more calls across the industry for stronger, coordinated safety measures.