The Information Security Awareness of Large Language Models