AI Researchers Developed a New “Jailbreak” Method to Test ChatGPT’s Defenses. Here’s What They Found.
A study pitted a specialized AI against common LLM chatbots like ChatGPT and Bard in an attempt to break down their defense mechanisms.Â
MJ Banias is an editor and co-founder of The Debrief. He covers science, technology and defense, and you can follow him at mjbanias.com and on Twitter @mjbanias.