We have always interacted with machines explicitly – typing, clicking, tapping. The next channel may be implicit and continuous: with wearables, EEG earbuds, and AR/VR headsets, the body's signals can be read to infer the mind, turning our mental state into the interface itself. This talk lays out that future honestly. Its core argument: nearly every hard problem in neural interaction traces back to one fact, that these signals are deeply individual. That cuts two ways. It makes the modelling brittle – noisy proxies, no ground truth, shifting baselines – and the data dangerously intimate, against the tightening backdrop of the EU AI Act and GDPR. Encouragingly, the fix is shared: separating common structure from individual deviation enables both personalisation and privacy-preserving, on-device modelling.