Security researchers discovered a new way to trick OpenAI’s language model, GPT-4o, into generating executable exploit code by leveraging a simple, yet cunning method—hex code. Hex-Encoded ...