Stylometric surveillance is already here and does not depend on the target telling the same story twice. I'm not even sure LLMs help the surveillors, but as a community we should investigate more into adversarial stylometry.