Cybersecurity researchers found it's easier than you'd think to get around the safety features preventing ChatGPT and other LLM chatbots from writing malware — you just have to play a game of ...