The fetch-files step in the ocrWF is should check to make sure that the files being fetched from preservation are copied properly to the OCR processing folder. This step copies files from the preservation system and routes them into ABBYY for OCR.
It looks like no check is being done currently because I recently had to manually fix a couple of items that were stuck in the ocrWF, and when I looked at the files, they turned out to be corrupt. The files in preservation were fine, so it looked like something went wrong when copies were made for OCR processing.