Talk about gone in 60 seconds

Computer scientists at the University of Maryland have developed an efficient way to craft prompts that elicit harmful responses from large language models (LLMs).…