Have not yet read the article, as Facebook is blocked at work, but I would guess that this is mostly the application of steganographic techniques, to hide known

Have not yet read the article, as Facebook is blocked at work, but I would guess that this is mostly the application of steganographic techniques, to hide known patterns, in datasets that are likely to be stolen/borrowed for training. Then observe the outputs of said models to try to discern related patterns.