Erez Yalon, head of security research at Checkmarx, spun up a shopping list program using a simple English-language prompt, then showed how easy it is to force an LLM to spew incorrect and dangerous results…
