Definitely some ML models today capable of identifying writers from their textual idiosyncracies...

Although I guess some adversarial code will extend it, the days of being truly "Anonymous" (and therefore liberated to speak on topics like this) are probably almost over.