AI-LADO, linguistic expertise, and the platformisation of asylum governance

Abstract The spread of Artificial Intelligence (AI) across public administration transforms how language and people are governed in Europe. These transformations are especially consequential in asylum procedures, where applicants have limited capacity to refuse technologies or contest their outputs. We conceptualise this reorganisation as the ‘platformisation’ of asylum governance and examine its consequences for linguistic expertise, evidentiary authority, and the political economy of language. We develop this argument through a study of AI-supported Language Analysis for the Determination of Origin (AI-LADO), focusing on Germany’s Language and Dialect Identification Assistance System (DIAS) and its circulation as a template for potential EU-wide implementation. Analysing institutional and policy materials, we show how DIAS transforms situated speech into file-ready probabilistic ‘indications’ presented as objective and auditable expertise. In doing so, it redistributes authority across caseworkers, technical systems, and commercial vendors and incorporates language into infrastructures of data extraction and reuse. Its reliance on licensed corporate software, commercial language-resource markets, and outsourced data labour also binds asylum administration to proprietary infrastructures. Rather than resolving the epistemic problems of human-led origin assessment, AI-LADO routinises and scales its reductionist assumptions. We situate this development within longer histories in which linguistic knowledge has served state classification and capitalist accumulation, while identifying platformisation as the contemporary form through which these relations are intensified.