Governing generative AI in multilingual higher education: a critical language policy analysis of Indonesian frameworks

Generative Artificial Intelligence (GenAI) education policies have proliferated across higher education institutions globally, yet their assumptions about the academic community they govern remain largely unexamined. Applying Bacchi's (2026. ‘What's the problem represented to be?': A new thinking paradigm. Routledge) ‘What's the Problem Represented to Be?’ (WPR) approach, this paper analyses GenAI education policy documents spanning two international, one national, and six institutional governance levels in Indonesian higher education. The aims are to examine how GenAI use is constructed as an object of governance and what assumptions underlie that construction. Across the seven Indonesian documents in the corpus, GenAI governance is consistently represented as a problem of academic integrity, ethical use, and institutional control. These framings share a common and unexamined presupposition: that the user subject of GenAI governance is monolingual, interacting with the tools in a single language and producing academic work in a single register. The analysis reveals that this assumption operates as a transfer filter. Despite one international guidance naming linguistic diversity as one of the first proposed policy measures, none of the seven retain this provision. The words bahasa daerah (regional languages) and multibahasa (multilingual) users or policy provisions appear in none of the Indonesian documents. Drawing on language ideology theory, this paper argues that the monolingual assumption is overdetermined, produced both by the selective transfer of Global North policy templates and by a pre-existing domestic standard language ideology arising from the standardisation of Bahasa Indonesia. This dual origin explains why the linguistic-diversity provision is consistently filtered out of the Indonesian documents. Moreover, the paper identifies the discursive, subjectification, and lived effects this assumption produces for multilingual GenAI users, and it proposes directions for linguistically inclusive governance in higher education.