You could post a hash of the paper to a trusted host like Twitter at the point you submit it, and you could then refer to that for proof of authorship and priority at a future date.
This idea would only work if everybody started securing some form of proof of authorship before sharing papers with anyone, which I don't quite see happening.