Non-breaking whitespace/other non-rendered characters can disrupt this type of technique. It can also be rendered moot by photograph, transcription, paraphrasin

Non-breaking whitespace/other non-rendered characters can disrupt this type of technique. It can also be rendered moot by photograph, transcription, paraphrasing, etc... Hell, even a re-encode/format change, or stripping of format info can moot the entire premise. I welcome Musk to demonstrating his own psychosis/neurosis. It doesn't really speak to any fundamental cleverness. Furthermore, were I a journalist receiving such material, I'd damn well make sure to change how I communicated it enough to neutralize attempts at such watermarking, simply because I know people are pre-disposed to such acts as he did. In fact, if you're going for maximum Discordian Malcontent high score, and you're confident you're operating under such scrutiny, neutralize a copy you leak, then carefully orchestrate the planting of false trails to keep the spook wannabes chasing their tails, which can end up blowing back on the orchestrator for waste of resources. This works better with a larger team though where Spartacus dynamics start kicking in. Sample size 1 tends to leave sufficient metadata from neutral parties to rehome in on a common perpetrator. This is why labor organization scares the living shit out of people like Musk, and other large holders of capital/abusive management types. Info asymmetry, fear, and disunity are some of their most potent weapons. Wisdom tends to prevail on the side of "don't do shit somebody'd find leak worthy, and you won't have this problem".