AI researcher David Kuszmar wrote this recent piece for IEEE Spectrum about his forays into the realm of llm guardrail-b...

AI researcher David Kuszmar wrote this recent piece for IEEE Spectrum about his forays into the realm of llm guardrail-busting techniques. During his adventures he managed to make the llm show him how to enrich uranium, create a meth lab, among other unsavory things. His technique involves setting up a scenario within a scenario, similar to how dreams are stacked up within dreams in the movie Inception. "How I Turned AI to the Dark Side"https://spectrum.ieee.org/jailbreaking-llms#ai #llms #security #hacking