Researchers from Italy discovered that phrasing prompts in poetry can be a reliable jailbreaking method for LLMs.