Austria's Court of Audit Reviews Federal AI Use — 35 Applications, One Written Risk Classification

The Austrian Court of Audit (Rechnungshof) examined the use of artificial intelligence in the federal administration between March and June 2024, covering mainly 2021–2023. The audit covered the Federal Chancellery, the Finance Ministry, the then Climate Action Ministry, the Ministry for Arts, Culture, Civil Service and Sport, and the federal IT provider BRZ GmbH. The report (series BUND 2025/20) was published in 2025.
Findings: the ministries reported 35 AI applications and projects as of June 2024 — 13 in production, 8 in implementation and 14 in feasibility testing. In the 20 application descriptions supplied, only one contained a written AI-specific risk classification, five used an AI-specific development or lifecycle model, eight took account of the federal 'Digital Administration and Ethics' guide, and none applied AI-specific certifications or standards.
The auditors also found that none of the ministries had an internal strategy explicitly addressing AI, that no ministry had a dedicated AI budget (AI costs sat inside wider IT projects), that no body had mandatory rules obliging staff to label AI use or AI-generated content, and that the 2021 national AI strategy's 64 measures did not assign responsibility to specific ministries or link to a budget. A 2023 inventory commissioned by the Finance Ministry was prepaid in full before delivery.
Recommendations include a standard procedure for AI projects with AI-specific risk assessment and standards, timely preparation for the EU AI Act, more training, and binding staff rules including transparent labelling of AI use. The audit is a cautionary, evidence-rich baseline: it documents governance gaps rather than measured service outcomes.

Read the full analysis: https://www.rechnungshof.gv.at/rh/home/news/Meldungen_2025/KI_in_der_Bundesverwaltung.html