AI Is Multilingual AI Truly Safe? Exposing the Vulnerabilities of Large Language Models in Low-Resource Languages