The vision of everyone having a personal superintelligent assistant is appealing, but it also means putting frontier capabilities in billions of hands, including bad actors. Is personal superintelligence the path to broad flourishing, or does wide deployment of powerful AI maximize misuse risk? Can both be true?