There are law enforcement projects underway to create databases of contraband files such as child porn or commercial software. These projects store information about the files without storing the files themselves, for the sake of faster queries as well as simply not wanting to store the stuff. The information is stored as hashes, or fingerprints, of each file. The simplest way to thwart such an automated database is to modify the illicit files in some way.

Fig. 5.1: Sample JPG file used.

Take an image (e.g. Fig. 5.1) and change it such that its hash value changes. The original file is called coffee1.jpg; the others are copies of this file, changed in one subtle way that won’t be noticed by the eye. The second file had the bottommost row of pixels cropped. The third file had its red channel color balance increased by 1. The fourth one was re-saved with the JPEG quality set to 99%.

As shown below, all four files have different hash values but contain essentially the same image -- making automated hash comparisons useless. It is simple to script these tasks, making their use over thousands of images trivial. The same thing could be done to illegally copied music files. Even pirated software can be disguised, by adding a small text file to its ISO.