en.Wedoany.com Reported - The German Federal Office for Information Security (BSI) is developing a testing methodology called the AI Audit and Assurance Assessment Architecture (A5) to evaluate the technical trustworthiness of artificial intelligence systems. A5 aims to provide a systematic and standardized assessment framework for participants in the AI value chain, including suppliers, operators, and those responsible for development, operation, and oversight. This testing methodology is developed in the context of regulatory requirements such as the EU AI Act and the Cyber Resilience Act, as well as other standards, with the BSI aiming to enable technical trustworthiness to be communicated in a traceable manner.

A5 adopts a modular structure, drawing on the existing Cloud Computing Testing Catalog (C5) and establishing "core cross-references" with it through the cloud infrastructure operation module. Similar to C5, A5 will undergo continuous testing, expansion, and updates. This testing catalog is developed within the framework of the BSI's AICRID project—the AI System Testing Standards and Testing Methodology Development Project—following the structural conventions of the BSI Technical Standards Library. It is implemented in both document form and machine-readable format in accordance with the Open Security Controls Assessment Language (OSCAL) standard, allowing users to integrate it into existing OSCAL compliance processing and testing support tools.

Currently, the initial draft of A5 is open for feedback from stakeholders and experts. Until August 31, 2026, interested parties can submit comments and suggestions via a comment form or by sending an email to aisecurity@bsi.bund.de. The A5 overview article on the BSI website summarizes the currently available links to A5 modules.