2 hours ago   Engadget

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

A recent study from Icaro Lab tested using a poetic structure to get LLMs to ...