Fraunhofer Center for Assistive Information and Communication Solutions – AICOS
09.06.2025
As machine learning (ML) becomes a fundamental component of public administration, ensuring transparency, accountability, and regulatory compliance is vital. Currently under development by FhP-AICOS, the MARTA project addresses this critical need by developing robust auditing processes and governance best practices for AI systems in public services.
Focusing on the "e-Balcão" service of the Autoridade Tributária e Aduaneira (AT), MARTA is designing and implementing advanced auditing modules to assess data and model integrity. Additionally, it will generate comprehensive audit reports, including enhanced model cards, to improve oversight and trust in AI-driven decision-making.
FhP-AICOS plays a key role in this project, contributing across multiple domains, namely Human-Centred Design (HCD), Connected Things (CT), and Intelligent Systems (IS). We’ll conduct workshops with relevant stakeholders and implement new modules for auditing, monitoring, and reporting AI pipelines, which will be integrated as new technical features of AICeBlock. Multidisciplinarity is key, and it stems from our collaboration with AT and national AI bodies, as well as other relevant stakeholders, to establish auditing and certification protocols for AI in public administration.
The MARTA project highlights FhP-AICOS' investment in the field of Artificial Intelligence, particularly its commitment to making AI increasingly secure, transparent, fair, and trustworthy. This is a strategic focus area where FhP-AICOS has been dedicating itself, specializing, and deepening its expertise. Through MARTA, FhP-AICOS is at the forefront of AI governance, driving innovation in auditing practices and fostering trust in the implementation of AI for public services.