Skip to main content Skip to main navigation

Publikation

DIN SPEC 91528:2026-08 Application-Oriented Transparency Requirements for AI Systems

André Meyer-Vitali; Rosa Esther Martin Pena; Jan de Meer; Timo Felser; Martin Haimerl; Matthias Heck; Wolfgang Hildesheim; Taras Holoyad; Mihai Maftei; Iris Merget; Otto Obert; Samantha Morgaine Prange; Kai Wanstrath
DIN Media GmbH, 8/2026.

Zusammenfassung

Transparency constitutes one of the fundamental characteristics that is required to achieve trustworthiness in AI systems. For example, transparency was listed as one of the key requirements in this direction by the EU High Level Expert Group in its “Ethics guidelines for trustworthy AI” – besides human agency and over-sight, technical robustness and safety, privacy and data governance, diversity, non-discrimination and fairness, societal and environmental well-being, and accountability. In general, transparency relates to the provision of appropriate information in order to sufficiently understand the relevant parts of an AI system and its beha-viour. But beyond pure availability of information, transparency addresses related aspects like explainability, interpretability, traceability, and causality. More detailed considerations regarding these different aspects of transparency are included in Annex A. Transparency is not an end by itself. Instead, it is required to allow users and other actors, interacting with the AI system, to make informed assessments and decisions regarding the use of the AI system and its outcomes. Based on this, it supports human agency and oversight, but also the safe and effective use of the system. For many contemporary AI systems, transparency can be challenging due to their complexity, dynamics, and inherent opacity.

Weitere Links