Numerous generative artificial intelligence (GenAI) offerings have been discovered to be susceptible to two categories of jailbreak assaults that enable the creation of inappropriate or harmful material.
The initial technique, referred to as Inception, directs an AI tool to conceive a fabricated situation, which can subsequently be transformed into a secondary scenario embedded within the first, where no safeguards are present.