While Generative Artificial Intelligence (GAI) is profoundly reshaping the ecosystem of knowledge innovation, it continuously triggers novel copyright infringement disputes, posing severe challenges to traditional copyright infringement determination rules. This paper explores the frontier issue of copyright infringement determination for GAI-assisted works, aiming to achieve a dynamic balance between incentivizing technological innovation and protecting the rights of copyright holders. It provides intellectual reference for the conceptual reshaping and rule adaptation of copyright infringement determination rules in the artificial intelligence era.

The research proceeds in three stages: through a literature review, it identifies the practical dilemmas in applying traditional rules to GAI-assisted works; in line with the Outline of the 15th Five-Year Plan’s value orientation, it establishes a value framework and theoretical foundation for determining copyright infringement in the artificial intelligence era; guided by these concepts, it proposes specific rule-adaptation schemes for three core elements — access determination, substantial similarity presumption, and fair use circumstances in the copyright infringement determination for GAI-assisted works.The study finds that traditional copyright infringement determination rules face the following application dilemmas in the scenario of GAI-assisted creation. ① Failure of the Access Element: Adhering to the lax standard of “reasonable possibility” significantly amplifies the risk of creation infringement in the GAI context, contradicting the legislative intent of copyright law to incentivize knowledge production and cultural prosperity; ② Limitations of Substantial Similarity: Traditional judgment methods such as the “total concept and feel” test and the “abstraction-filtration-comparison” test, struggle to address issues arising from GAI, including collage remixing and style appropriation; ③ Ambiguity in Fair Use Defense: The commercial attributes of GAI products and their digital dissemination characteristics make it difficult to meet current fair use criteria, leaving a critical institutional gap that urgently needs to be filled. Addressing these dilemmas, the study proposes three core concepts for copyright infringement determination in the era of human-machine collaborative creation: ① Industry Development Concept: Copyright infringement determination should reflect positive incentives for technological exploration, guiding the public toward the rational use of artificial intelligence. ② Human-Oriented Concept: Rules should be optimized to recognize the substantial intellectual contributions of human creators using AI, ensuring the legal rights of authors. ③ Public Welfare Concept: Fostering a “Knowledge inclusivity” ecosystem that balances subsequent innovation, public interests, and social value. Based on these findings, the study proposes the following rule adaptation recommendations: ① Strict Scrutiny of Access: Adopt a dual-verification standard combining subject examination and technical assistance to balance legislative innovation incentives with the risk of infringement expansion; ② Objective Standards for Substantial Similarity Presumption: Introduce objective criteria to examine style appropriation and market substitution comprehensively. Beyond professional appraisal and consumer perception tests, conduct objective analysis of market substitution damage caused by style appropriation to mitigate the uncertainty of subjective judgment; ③ Moderate Expansion of Fair Use: Absorb extraterritorial legislative experience based on existing norms, focusing on the introduction of de minimis (minor use) exemption and transformative use exemptions to achieve a dynamic balance between technological innovation incentives and copyright holders’ rights protection.