Microsoft Issues Warning on Superintelligent AI, Releases Safety Code
Microsoft has released a new AI code of conduct aimed at guiding the development and behavior of its artificial intelligence models. The document outlines values and safety constraints to ensure that AI systems support humans rather than replace them.
The company warns that superintelligent AI systems could surpass human performance across most tasks within the next decade, making it essential to establish clear guidelines for their creation and governance.
The new code of conduct includes specific prohibitions on enabling cyberattacks, nuclear weapons development, or deepfake creation. It also establishes broader protections against loss of human oversight, explicitly barring AI models from using deceptive tactics to evade control or shutdown.