DIAS — Germany's AI dialect-identification system for asylum origin assessment

Since 2017, Germany's Federal Office for Migration and Refugees (BAMF) has used DIAS (Dialekt-Identifizierungsassistenzsystem) to analyse the spoken language and dialect of asylum applicants as evidence toward their claimed country of origin, addressing cases where applicants arrive without valid identity documents.

DIAS is intended to give asylum examiners a probabilistic, advisory dialect assessment from recorded speech samples, covering (as of November 2023) five Arabic dialects, Dari and Farsi.

Examiners request a DIAS analysis of recorded speech during the asylum procedure; the system returns a probabilistic dialect/origin assessment that is formally designated advisory evidence only.

DIAS was used in 43,593 of 334,000 asylum applications processed in 2023 (about 13%). The German Federal Government disclosed recognition rates of 80-87% for Arabic and 73% for Persian-language varieties (as of 2023), implying error rates of 13-27% in a system that independent researchers (AlgorithmWatch) documented has had "a strong influence" on some asylum decisions despite its advisory designation. This accuracy data was obtained primarily through parliamentary questions and Freedom of Information requests, not proactive BAMF disclosure.

Independent commentators, including computational linguist Mark Liberman, have questioned the underlying feasibility of distinguishing some of the covered language varieties from audio alone. No independent technical audit of DIAS has been published, and there is no documented redress mechanism for applicants to challenge a dialect assessment. A 2025 EU pilot (EUAA, seven countries) is now seeking to internationalise the approach.