Recent content by noahspar

A useful way to understand AI Security is to separate the goal, the input, the model or tool, and the way the output is evaluated. One important area is security. Define what a good result looks like before changing tools or settings, then test one variable at a time so you can tell what...