It's also because it's brand new.

Give it a few weeks for them to classify its outputs, and they won't have a problem.

I’m less sure. This seems like an asymmetrical battle with a lot more money flowing to develop the models that write than detect.