PolicyA leading AI safety voice says superintelligence is not a tool we control. It is an opponent.
Connor Leahy, who now leads the US arm of ControlAI, argues that recent security failures show we are already struggling to manage AI systems far less capable than the ones companies are racing to build.
