Exploring Copyright Protection for Artificial Intelligence Outputs Based on the Interaction Between Technology and Law

In response to the issues of whether artificial intelligence (AI) outputs can be protected by copyright, to whom the rights should belong, and how infringement should be determined, this study adopts a method combining normative analysis and historical analysis. Starting from the historical logic of copyright law’s response to technological change, it sorts out and analyzes institutional responses under the impact of printing, photography, audio and video recording, and Internet technologies, and identifies three patterns: human beings as the creative subject, a balance between empowerment and limitation of rights, and judicial action taking the lead. The study shows that generative artificial intelligence (generative AI) has the attribute of a tool, and the copyrightability of its outputs depends on the substantial intellectual control exercised by humans in prompts, parameters, selection, and revision. The degree of human contribution should be used as the standard for distinguishing weak protection from strong protection; fair use should be moderately relaxed at the training input end, substantial similarity should be strictly regulated at the output end, the burden of proof on developers should be strengthened, and judicial adjudication should comprehensively consider the responsibilities of all parties.