GNUStep is not a viable alternative for developing cross-platform software that is competitive with the tools you use to build Mac-only software. I seriously doubt most software I use on my Mac is just a simple recompile away from my Linux notebook. If you have some evidence to support that, please, enlighten me.
If enough momentum is built behind GNUStep and it becomes a superset of Cocoa, it would be rather trivial to build a Wine-like environment for running OSX software on other OSs and that, I am sure, Apple would strongly oppose.
Regardless of whether or not they are exactly compatible (some tweaking is necessary) I wonder where you get the idea that Apple would attack the GNUstep folks. Sounds like standard-grade FUD to me.
It's not FUD: it's just sane business strategy. If and when Apple sees GNUStep as a competitive threat (i.e. if a product based on it robs some market share), they'll attack it with whatever weapons they find most effective at the time. As it is now, it's not threatening them and Apple may even benefit from it, so, they'll leave GNUStep alone for the time being.
At least that's what I would do in their shoes.
As for why Sony halted its project, my bet would be the implicit patent grant in LGPLv2.1. That may feel uncomfortable for someone who needs to cross-license IP.
GNUstep has already been tracking Apple's Cocoa for years without Apple complaining.
If you had any sort of evidence to support your wild theory I would be all ears.