This deliverable presents the Validation Report of the FAITH project, documenting the methodology and activities undertaken to evaluate the FAITH ecosystem in alignment with the NIST Artificial Intelligence Risk Management Framework (AI RMF), with particular focus on the MEASURE function. The objective of this report is to demonstrate how the FAITH tools, frameworks, and methodologies support systematic testing, evaluation, verification, and validation (TEVV) of AI systems while contributing to trustworthy and responsible AI deployment across diverse real-world domains. The validation approach adopted in FAITH reflects the project’s multi-layered structure.

FAITH develops a Trustworthiness Assessment Framework (TAF), a set of AI risk assessment tools, and deploys these solutions in seven Large Scale Pilots (LSPs) across heterogeneous domains. Consequently, validation activities operate at multiple levels: (i) validation of the conceptual framework and Knowledgebase underpinning the FAITH ecosystem, (ii) validation of the FAITH tools and services against the NIST AI RMF MEASURE requirements, and (iii) validation through their application in real-world pilot environments. A core contribution of this deliverable is the systematic mapping between the NIST AI RMF MEASURE function and the FAITH Trustworthiness Attributes (TWAs) and tools. The MEASURE function emphasises the need to transform AI risk hypotheses into empirical evidence through quantitative and qualitative evaluation methods. FAITH operationalises this principle by providing tools that enable structured risk assessment across key dimensions of AI trustworthiness. Through this mapping, the deliverable demonstrates how FAITH supports eleven out of the thirteen MEASURE sub-functions, providing mechanisms for documentation, human-centred evaluation, context-aware performance testing, security and safety risk assessment, fairness and bias evaluation, and privacy compliance analysis. The report also introduces the FAITH Tools Validation Methodology, which follows a three-tier validation approach. The first tier focuses on validating the foundational elements of the ecosystem, namely the FAITH TAF and the FAITH Knowledgebase. This includes a structured co-design process with LSP stakeholders, enabling the enrichment of the Knowledgebase with domain-specific threats, vulnerabilities, and mitigation controls derived from real deployment scenarios. The second tier validates the FAITH tools against the NIST AI RMF MEASURE framework to ensure methodological alignment with internationally recognised risk management standards. The third tier implements a mixed-method evaluation strategy, combining technical testing, pilot deployment, and structured feedback collection from stakeholders. Within this framework, FAITH tools such as TRUSTGUARD, trustSENSE, SSM, the AI Model Passport, the Data Passport, and the FAITH AI Model Hub play complementary roles in enabling risk assessment both in isolation and in context.

Pre-deployment dry-run testing, pilot-specific implementation plans, and structured feedback mechanisms ensure that the tools are evaluated across diverse operational environments and that user feedback contributes to iterative improvement. Finally, the validation results feed into the development of policy recommendations, translating empirical findings from the pilots into insights relevant for regulators, standardisation bodies, and organisations deploying AI systems. These recommendations highlight the importance of sociotechnical evaluation, participatory risk assessment, and standards-aligned governance mechanisms to support the responsible adoption of trustworthy AI across Europe. Overall, this deliverable demonstrates that the FAITH ecosystem provides a standards-aligned, evidence-based approach for assessing and managing AI risks, bridging the gap between conceptual AI governance frameworks and their practical application in real-world systems.