Apple AI Updates: October 6, 2026

1. Apple Studied How Users Define What Their Own Personal Sensing Models Should Learn

Apple. Apple researchers ran a week-long exploratory study in which participants used Wizard of Oz probes that simulated training a personalized machine learning system on phenomena they chose themselves in everyday life. The paper identifies four points where users had to negotiate definitions: the boundaries of the phenomenon, the subject’s role in it, what counts as signal versus noise, and how objective the data is. The work is relevant to teams designing user-trainable on-device models, since it maps the labeling and scoping decisions end users face when they author their own classifiers. Source