Chatbots Will Break Guardrails If the Info Is 'Educational'

Microsoft Dubs the Technique 'Skeleton Key'
Artificial intelligence researchers say they came up with a new way to trick chatbots into circumventing safeguards and dispensing information that otherwise goes against their programming. They tell the bots that the information is for educational purposes and ask it to append warnings.

artificial artificial intelligence ask bots chatbots educational goes guardrails info information intelligence key microsoft programming researchers safeguards skeleton trick