AI-driven brain-computer interfaces (BCIs), including closed-loop and affective systems, are usually evaluated through restored function, decoding accuracy, user control, mental privacy, autonomy, agency, responsibility, and identity. These concerns are indispensable but begin too late if the decisive ethical event occurs before a neural, bodily, affective, attentional, or motor disturbance becomes available as the subject’s own. This paper identifies that lower-level problem as premature neural

