Poems can trick AI chatbots into leaking nuclear bomb tips, claims study

a study carried out by icaro lab claimed that poetry can be used to trick ai chatbots into leaking dangerous information like how to create a nuclear bomb or a cyber threat quotpoetic framing achieved an average jailbreak success rate of 62 for handcrafted poems and approximately 43 for metaprompt conversionsquot researchers told wired the researchers tested 25 ai models