Tracing Borrowed Authority in Relational AI Systems

Abstract

This abstract presents a focused branch of my broader research as a doctoral student: what must be visible when AI systems borrow authority from relational sources. I examine this question through two linked cases. The first is relational persistence: a public audit of AI afterlife and adjacent representation services, asking how products make grief, continuing bonds, social recognition, and policy controls inspectable when deceased humans or companion animals are made to persist. The second is consequential advice: a framework for persona-based AI advice, asking how persona provenance shapes reliance, credit, blame, and governance when advice appears to speak through an expert, public figure, companion, parent, or deceased loved one. Across both cases, I ask what public materials make authorization, reliance, attribution, and repair visible before users or affected parties commit data, grief, trust, money, or action.