GenAI as Fictional Author: Statutory Attribution of Copyright in AIGC

Abstract The capacity of GenAI to produce content autonomously and at scale challenges copyright doctrine because the machine-generated portions of AIGC may satisfy originality thresholds in a factual sense while lacking a qualified legal author. This tension is especially visible in the divergence between the United States, which continues to insist on human authorship, and China, where some courts have been more willing to protect AI-related outputs through human attribution or pragmatic reasoning. A further doctrinal distinction is required between the copyrightability of user inputs, including prompts and subsequent edits, and the copyrightability and ownership of the final AI-generated output. Against this background, two routes are examined. The first treats GenAI as an electronic person or juridical person for limited attributional purposes, but it is ultimately rejected because it currently faces substantial ethical and institutional barriers. The proposed framework instead treats GenAI as the fictional factual author of the machine-generated portions of AIGC and attributes legal authorship, as a matter of statutory allocation, to the developer or other controlling deployer, while preserving user rights in copyrightable prompts, selection, arrangement, or revision. This framework seeks to clarify authorship, ownership, and accountability without requiring full legal personhood for AI.