THE BOUNDARY BETWEEN AVATAR AND HUMAN: AUTHENTICITY CHALLENGES IN DIGITAL PERSONALITIES

Abstract

AI systems can now replicate how a specific person spoke, thought, and made decisions. Voice cloning passes human perceptual tests. Large language models reproduce individual vocabulary and rhetorical style from a text corpus. Brain-computer interfaces are beginning to map the neural activity behind personality traits and decision-making habits. What was once a philosophical thought experiment is becoming an engineering project: building a digital version of a person that can hold a conversation, respond to new questions, and act on its own. But as these digital personas grow from stored recordings into adaptive systems that learn and change over time, two questions become unavoidable. How do we tell the difference between a faithful copy and something that has become its own entity? And who is responsible when a digital persona says or does something its human original never approved? This article works through the philosophical, ethical, and legal sides of that problem. Drawing on theories of consciousness, digital identity law, and hands-on experience building AI persona systems, we propose a five-level classification of digital persona autonomy (from Static Record to Digital Consciousness) and introduce the Authenticity-Accountability Matrix (AAM), a model that links each autonomy level to a specific accountability structure and set of authenticity criteria. The classification and matrix are grounded in a case study of OPV Systems’ “Last Call” platform, a grief support application that reconstructs the communicative identity of deceased individuals for their families. Current legal and ethical instruments cover digital personas only at the lowest autonomy levels, leaving a regulatory gap for adaptive and autonomous systems that will demand new governance approaches built across disciplinary lines.