Stop Googling. The answer is staring you right in the face—you just have to read it.

These new models are specially trained to recognize when an LLM is potentially going off the rails. If they don’t like how an interaction is going, they have the power to stop it. Of course, every ...