Abstract
Political science education in Southeast Asia faces two related challenges. Curricula remain dominated by Western-centred theoretical frameworks that inadequately reflect the historical and institutional realities of the Association of Southeast Asian Nations (ASEAN), and artificial intelligence (AI) tools used in education are trained predominantly on Global North data, risking the reproduction of epistemic dependencies. This conceptual study pursues three interrelated objectives: examining the current state of AI integration in ASEAN political science education, analysing how regional AI governance frameworks conceptualise the relationship between AI, education, and sovereignty, and proposing a structured pedagogical response. A qualitative research design combining document analysis and comparative policy analysis was adopted. Twelve primary policy and governance documents, supplemented by peer-reviewed literature published between 2021 and 2026, were analysed through a three-stage thematic coding procedure. The findings identify an epistemic dependency problem comprising four dimensions: theoretical, technological, pedagogical, and assessment dependency. The Discussion explains why this dependency arises and proposes the Sovereign AI Pedagogy Framework for Political Science (SAPF-PS), an integrated model comprising four interdependent pillars: Epistemic Grounding, AI-Enhanced Methods, Sovereign Assessment Design, and Critical Digital Literacy. As the findings derive from interpretive document analysis, the SAPF-PS is presented as a conceptual contribution pending future empirical validation. The study also considers the risk that AI sovereignty rhetoric may be appropriated to justify state control over information rather than to strengthen educational autonomy, and argues that this risk must be actively managed rather than assumed away. The study concludes that embedding AI sovereignty within curriculum design, classroom activities, and assessment practices may strengthen ASEAN's capacity to shape its own AI-informed political futures.