Regarding the debate over whether 7601.17514 is the real RTM or if it's just another RTM escrow (i.e., a RTM candidate that was not signed-off on), it seems that everyone and every "news" source that people have pointed to points back to that Russian TechNet blog post. It's really quite amusing to read some of the posts in this thread, where someone would point to the TechNet post and then, as an extra source, to some "news" article that was ultimately sourced from that same TechNet post.
In all this, I find it surprising that nobody noticed the biggest and most convincing piece of evidence in support of 7601.17514 being the signed-off RTM. Take a close look at KB976902 (the servicing stack update; support.microsoft.com/kb/976902). You will notice that a number of the files found inside this update are 7601.17514 (the others carry an original 7600.16385 build number). This is far more convincing to me than some Russian TechNet blog post (unless it came from Redmond, I don't really trust TechNet blog posts) or the "have faith in wzor" doctrine. The fact of the matter is, Microsoft has already public released some of the bits in 7601.17514 via WU, and if that isn't a signal of sign-off, I don't know what is!
As for how this leaked, I suspect that it was probably "leaked" in the usual manner: Microsoft has a long history of supplying their close partners (esp. the "royalty" hardware partners, e.g., Dell) with early bits and frequent iterations of betas, RCs, etc. Redmond's security is actually pretty tight; it's almost always the third parties involved that are the sieves.