The algorithmic autonomy and data-driven characteristics of generative artificial intelligence are reshaping content production paradigms. Deep learning models, trained on massive datasets, can continuously generate content without human intervention, breaking away from the 'human author-centric logic. The current legal framework faces challenges in terms of subject qualification, as AI-generated content fails to meet the criteria for rights holders and struggles to establish reasonable presumptions of rights ownership through existing systems. Regarding originality assessment standards, traditional criteria struggle to evaluate the expressive differences in algorithmically generated content, and the black-box nature of the creation process leads to a 'quantification dilemma in determining originality. The technological impact also manifests in the dimension of rights allocation. The legality of training data copyright, the rights and responsibilities of secondary dissemination of generated content, and the lack of regulations on neighbouring rights systems fail to address diverse interest claims. The current legal framework's lag has triggered systemic risks, failing to incentivise innovation or balance value conflicts, necessitating the urgent establishment of a new governance paradigm.
F. Qi (Wed,) studied this question.