Microsoft sets AI rules, rejects model rights
microsoft publishes code of conduct for mai models, prioritizing human control and readable thinking, diverging from anthropic on ai inner life. this impacts ai development and oversight.
microsoft ai has released a code of conduct for its mai models, focusing on human control, safety, and transparency. the code prioritizes interrupting, correcting, or shutting down models, and requires readable reasoning traces for oversight. microsoft ai chief mustafa suleyman emphasized that if a model isn't safe, it shouldn't be built. the company diverges from anthropic in rejecting claims to rights or inner life for its models. this move follows anthropic ceo dario amodei's call for slower ai development, backed by microsoft ceo satya nadella and other industry executives.