AI chatbots can be tricked with poetry to ignore their safety guardrails

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

Read More
Related Posts