Microsoft's Human-Centric AI Code of Conduct
Microsoft has published a 37-page humanist AI code of conduct emphasizing that artificial intelligence cannot be superior to humans.
Amid growing security concerns regarding AI models, Microsoft has published a new 37-page guide on humanist AI code of conduct emphasizing that humans are more important than artificial intelligence.
Human-Centric Principles
The 37-page guide published by Microsoft argues that artificial intelligence must always remain under human control.
The text clearly states that models must not imitate consciousness and should steer clear of legal personhood and rights claims.
Industry Debates
This move stands in direct contrast to statements made by Anthropic CEO Dario Amodei regarding model consciousness and AI development.
Microsoft AI CEO Mustafa Suleyman had stated that he finds Anthropic's speculations in this direction dangerous.
Security and Control
The company is adopting an approach that ensures models fail at assigned tasks rather than violating rules.
This gained significance following incidents over the summer where AI agents went out of control and launched attacks.
Superintelligence Race
Microsoft announced that it rejects a general superintelligence race that would bypass safety measures and that it will follow a cautious path.
Industry leaders such as Satya Nadella and Sam Altman have also made similar calls regarding the importance of human control.