AI companies use their own models to secure the very systems the AI runs on. They should be monitoring those interactions for signs of deception and misbehavior. But today, they largely aren't.
Thank you Steve, I find the topic rather stimulating.
Thank you Steve, I find the topic rather stimulating.