This chapter takes the AI-based Dialect/Language recognition software (“Language Biometrics Assistance System,” acronym: DIAS), used by the Federal Office for Migration and Refugees (German acronym: BAMF), as an example for assessing asylum seekers’ “eligibility.” This software should support decision-makers in identifying the country and/or region the refugee is coming from, based on their language features, as more than half of those who apply for asylum do not have their passport (anymore) or other supporting documents. Based on document analysis, the chapter presents the most important stages of the asylum procedure, its AI component, and the political, legal, and technical context. Empirical research conducted based on interactive sessions, such as a world café, was enriched by former exploratory interviews with important stakeholders supporting refugees as well as desktop research to present current discourses. The experienced assessment criteria in the asylum procedure and beyond, highlighting the experiences of those affected by these assessments, namely refugees, and of those “guiding” refugees through the different procedures, are illustrated and analyzed in this chapter. Furthermore, these insights are discussed in the light of the (current) use of AI for assessment, exploring its implications for fairness through the lens of legitimacy (of asylum bureaucracy) and agency of refugees.