But that's a bizarre argument. Many uses other than encryption might also do this (lets say, hypothetically, that it was indeed a Public Radio Entropy Source)
So, we're going to take an empty channel, filled with random noise, and replace it with a transmission filled with random noise! Which will be less random than what we started with.
I'm not impressed yet
Go on believing that steganography can't be detected. I'd rather be able to watch you, if necessary, than not.