New Guidance for Risk Management of Artificial Intelligence Systems

The European Data Protection Supervisor (EDPS) is pleased to announce the publication of a new guidance document designed to support controllers in conducting data protection risk assessments when developing, procuring, and deploying Artificial Intelligence (AI) systems under Regulation 2018/1725 (EUDPR). This guide aims at providing valuable insights and practical recommendations to help identify and mitigate common technical risks associated with AI systems, helping in the protection of personal data.

While primarily intended for European Union Institutions, Bodies, Offices, and Agencies (EUIs), this guidance is also relevant and useful for private companies, industry stakeholders, and public organizations seeking to ensure compliance with data protection regulations.

The document begins by revisiting the risk management approach of the widely recognized ISO 31000:2018 standard. It then continues into the AI system lifecycle, to later explore the concepts of interpretability and explainability, which are essential for ensuring data protection. The core of the guidance presents a detailed analysis of risks and corresponding mitigation measures, organized around four fundamental data protection principles: fairness, accuracy, data minimisation, and security.