Overview
On 20 October 2026, the consortium of the research project “VERITAS” convened for a one-day workshop at the Austrian Institute of Technology (AIT) in Vienna to officially kick off the project and begin conceptual work. The Research Institute was represented by Robert Rothmann and Madeleine Müller. Funded under the KIRAS Security Research Programme, the project focuses on developing transparent and explainable AI-based assistance for complex procedures in the asylum system.
Project Background and Objectives
VERITAS stands for “VERtrauenswürdige Identitätsfeststellung durch Transparente Agenten-Systeme” (Trustworthy Identity Verification through Transparent Agent Systems). The project addresses the demanding work carried out by the Austrian Federal Office for Immigration and Asylum (BFA), which involves evaluating identity documents, interview transcripts, country reports and existing case files, and cross-referencing them to assess consistency. Examining this extensive information requires particular care and involves considerable time and cognitive effort.
This is where VERITAS comes in with the development of the AI assistance system “ALIGN”. Designed as an interactive copilot, the system is intended to pre-structure case files and flag potential spatial and temporal inconsistencies. Its primary focus is on supporting identity verification and ensuring the quality of administrative decisions by cross-checking information from different sources.
System Architecture and Legal Requirements by Design
From a technical perspective, VERITAS employs a hybrid neuro-symbolic approach that combines an open multi-agent architecture with logic-based verification methods. Information is extracted from documents, consolidated into a structured knowledge graph and checked for inconsistencies against predefined rules. This approach is intended to make results verifiable and understandable to users.
“Compliance & Ethics by Design” is a central component of the project: legal and ethical requirements are to be incorporated into the system architecture from the outset. Explainable AI and the systematic integration of human review are intended to counteract uncritical reliance on automated results. The planned deployment on local hardware also supports data sovereignty.
Multidisciplinary Consortium and Outlook
Led by AIT, VERITAS brings together the Austrian Federal Ministry of the Interior, the Research Institute and X-Net Services GmbH. The project runs from October 2026 to September 2028. In addition to a validated software prototype, the long-term objective is to develop a modular open-source framework that could serve as a foundation for the use of AI in European public administration in accordance with the rule of law.
The Research Institute is pleased to be part of this important research project and to contribute its expertise at the intersection of data protection law, artificial intelligence and ethics. Together with our project partners, we aim to advance research into trustworthy AI systems and help ensure compliance with legal requirements from the very beginning.