Exploring Attribution Biases in Responsibility Assignment during Digital Human Live Streaming: An Experimental Analysis

Abstract

Digital human technology has gained extensive adoption in live streaming commerce, with virtual anchors demonstrating core value in brand promotion and e-commerce monetization. Through controlled experiments examining the effects of anchor positioning, event attributes, and anthropomorphism levels on audience responsibility attribution, findings indicate significantly heightened bias toward blaming virtual anchors during negative events—particularly under high anthropomorphism conditions. This empirical evidence substantiates the necessity for ethical guidelines and regulatory frameworks in digital human applications. As industry standards mature, designing a rational responsibility-attribution framework for AI streamers has emerged as a critical imperative to reconcile technological innovation with consumer protection.