The AI risk problem has already been solved. If any robots start acting up, just hit 'em with the ole semicolon and down they go.