AI safety tests reveal chatbots provide dangerous instructions like bomb-making and bio-weapon details, raising major security concerns.....