831 mitigation actions extracted from 20 AI governance and standards frameworks, organized into a preliminary taxonomy of 4 categories and 28 subcategories.
The database catalogues concrete mitigation actions — specific things actors can do to reduce AI risk — drawn from established governance and standards frameworks including the NIST AI RMF, EU AI Act, and ISO/IEC standards. Each action is classified into one of 4 control categories: governance and oversight, technical and security, operational process, and transparency and accountability, with 28 more granular subcategories.
Identify what interventions exist for specific risk areas, compare how different frameworks approach the same problem, and build out risk assessments or compliance checklists. Each mitigation links back to its source framework for detailed implementation guidance.
How mitigation actions distribute across subcategories, broken down by source framework and control category.
A hierarchical view of the mitigation taxonomy showing 4 categories and 28 subcategories with their relative sizes.
Browse all mitigation actions or use filters to narrow by category, subcategory, source framework, or keyword.