It's not a code embedded into CP, it's a hash of the image/video content. So what you'd want to do is find out one or more of the hashes (legal to possess, it's just an alphanumeric identifier) then create an image/video that causes a "hash collision" (i.e., same output hash but different content). This is easier said than done but is perfectly possible, especially by using GPU for highly-distributed hash calculations.