The rapid advancement of generative artificial intelligence (AI) has posed significant challenges to the legal framework governing intellectual property (IP) protection, particularly with respect to the collection and use of data for training AI models. Contemporary AI systems are typically trained on massive datasets, many of which contain copyright protected works. This practice raises potential risks of intellectual property infringement and underscores the need for an appropriate legal framework that balances the protection of the legitimate rights and interests of creators with the promotion of technological innovation. This article employs a combination of research methods, including document analysis, synthesis, and comparative legal analysis, to examine China’s legal framework governing intellectual property protection in relation to the use of AI training data. These mechanisms encompass: (i) copyright-based protection; (ii) administrative management and data transparency; and (iii) the legal liability regime applicable to AI service providers. The article thereby proposes policy recommendations for improving Vietnam’s legal framework, with a view to fostering a transparent, equitable, and sustainable AI ecosystem.