This study examines shifting patterns in global academic knowledge production through the lens of world-systems theory, focusing on the role of generative artificial intelligence (AI) in reshaping epistemic hierarchies. Drawing on longitudinal bibliometric data (1996–2023), it analyses publication volumes, international collaborations and open access trends across core (U.S., U.K.), semi-peripheral and peripheral regions, with special attention to China’s emergence as a leading scientific producer. The paper highlights a growing divergence in AI infrastructure and access: Western scholars increasingly rely on open tools like ChatGPT, while China’s closed ecosystem is governed by distinct political and epistemic norms. Thesedevelopments may entrench parallel scientific systems, exacerbating inequalities in visibility, legitimacy and collaboration. As AI becomes a central driver of researchpractices, this paper argues for inclusive, interoperable knowledge infrastructures to avoid deepening global academic fragmentation. The findings offer a critical perspective on the geopolitics of knowledge in the digital age.