When personal models trained over a lifetime become the most intimate assets in an estate, families must decide whether a dead person's learned digital self should be erased, archived, or allowed to advise the living.
Personal models accumulate decades of private conversations, preferences, and unfinished plans. As they become more useful than generic cloud models, inheritance law expands beyond files to cover learned patterns of behavior. Model wills let people specify who may consult, retrain, or delete their digital counterpart, creating a new ritual around death and memory.
At 10:15 on a rainy Tuesday morning in a Busan probate office, Mina places her father's old tablet beside his paper will. She allows the model to answer only three questions about the family workshop, then watches the clerk permanently disable its ability to reproduce his voice.
A model's output is not the deceased person's judgment, and granting it special legal status could turn probabilistic imitation into false authority. Wealthy families might preserve sophisticated ancestral advisers while poorer households lose their digital histories when unsupported devices fail.