Roadmap for AI Act Risk Level Classification

This service supports organisations in understanding and navigating the risk assessment process for AI systems under the AI Act. It provides a clear preliminary risk classification together with an explanation of the associated regulatory implications. In addition, it identifies gaps in compliance with AI Act requirements and provides practical guidance on how to address high-risk systems. The service also delivers a structured testing and compliance roadmap to reduce risk and speed up market readiness, and offers clear input for further expert legal consultation when classification is uncertain or complex.

The service consists of four steps. First, information collection, where the customer provides use-case details together with technical and functional descriptions, deployment context, and operating environment, based on a questionnaire prepared by RISE. Second, a face-to-face or online workshop involving both RISE and the customer is conducted to clarify details and close information gaps. Third, the risk classification analysis is performed by RISE. Finally, the results are presented in a concluding workshop, where the risk classification report summarising outcomes, rationale, and recommendations for strategic next steps is presented. The report may include a list of suitable testing methods and potential adaptations to reduce the risk class. Service execution typically takes 3–4 weeks, but can be changed to suit the customer. Deliverables from the service include a Preliminary Risk Classification Report, a completed Questionnaire and Evidence Record, and a Compliance Gap Table highlighting missing or partial measures required under the EU AI Act.

Customization: The service can be tailored to the customer’s sector, deployment context, and roles by adjusting the scope, depth, and focus of the analysis while applying the same core AI Act classification methodology. Limitations: The results are preliminary and guidance-based, and their reliability depends on having sufficient and accurate documentation about the AI system, its intended use, and its deployment context.