Perhaps someone with more expertise of Soviet archives as a whole might want to jump in, but from my more limited experience with them, I don't think this specific concern is a huge problem, although there may be isolated instances. For one, we're talking about declassified materials - ie stuff that was not made public during the Soviet era - which would raise the question of what the point of doctoring them would be if no one was expected to read them. Remember, the Soviet Union didn't expect to collapse and for a new government to suddenly let researchers access huge numbers of secret files. That's on top of the sheer scale of these collections - editing millions and millions of documents en masse is a huge task, let alone doing so convincingly.

That said, as with any archive but perhaps particularly Soviet archives, we can't treat any material as completely reliable. These documents were created, read and stored in a historical context with a specific purpose, all of which influenced how they were composed and selected for preservation. Officials would likely have known not to write certain things down or express certain ideas, and may themselves not have known the 'truth' in any case. Navigating these constraints, and assessing the usefulness of individual bits of evidence in light of them, is one of the more fundamental skills of historical research.