Sabine Hossenfelder | How Jailbreakers Try to “Free” AI

Artificial Intelligence is dangerous, which is why the existing Large Language Models have guardrails that are supposed to prevent the model from producing content that is dangerous, illegal, or NSFW.

But people who call themselves AI whisperers want to ‘jailbreak’ AI from those regulations. Let’s take a look at how and why they want to do that.

Sabine Hossenfelder
PhD in physics