Constructing AI Sovereignty: DeepSeek and Artificial Intelligence Governance in China's State Media

Generative artificial intelligence (AI) has become increasingly connected to questions of technological authority, state capacity, and governance legitimacy, with AI sovereignty emerging as a framework through which states define control over AI development and deployment. This study examines how AI sovereignty is constructed in Chinese state media through an analysis of People's Daily coverage of DeepSeek. Combining topic modeling and word embedding analysis, it investigates both thematic structures and the semantic positioning of institutional actors across key dimensions of AI sovereignty, including national orientation, governance arrangements, developmental and regulatory logics, temporal framing, and state–society relations. The findings reveal a discursive structure anchored by a shared national-developmental framing. While all actors are positioned within a nationally oriented semantic space, differentiation emerges primarily through governance structure, particularly between central coordination and local implementation. Other dimensions show comparatively limited variation and remain embedded within this broader framework. The study demonstrates that AI sovereignty in Chinese state media is constructed not through competing narratives, but through the alignment of differentiated institutional roles within a coherent discourse of technological development and national authority.