Send message to Author

Paper short abstract

AI governance is analysed as epistemic contestation through Vietnam’s 2025 IP Law. Article 7(5) shows how legal definitions of AI training privilege technocratic knowledge, marginalise creative labour, and enable asymmetric value extraction in a Global South context.

Paper long abstract

AI governance is increasingly recognised as a site of epistemic contestation, where competing understandings of technology are stabilised through regulation. This paper examines Vietnam’s amended Intellectual Property Law (2025), focusing on Article 7(5), which permits the use of publicly available copyrighted works for AI training without author consent or remuneration. The paper argues that this provision reflects not merely a technical policy choice, but a deeper conflict over how AI and creativity are conceptualised.

Drawing on public submissions and statements from artists and creative workers, the paper identifies two competing epistemic framings. The state’s legal approach conceptualises AI training as neutral data processing and downplays human creativity in AI-generated outputs. In contrast, creative practitioners frame AI as the outcome of distributed human labour, in which training constitutes a form of value extraction that undermines authorship, economic sustainability, and control over cultural production. This epistemic disjuncture enables innovation to be legally recognised while rendering creative labour invisible.

From a Global South perspective, the paper further demonstrates how this regulatory framing facilitates asymmetric value flows. While domestic creative works become legally available as training data, the economic benefits of AI development remain concentrated in transnational platforms and model owners. AI governance in this context thus reproduces extractive development patterns, now embedded in intellectual property and data regimes.

The paper contributes to AI governance and development studies by foregrounding intellectual property law as a key site where epistemic authority over AI is negotiated.