The AI Risk Repository provides a publicly accessible, continuously updated database of 777 AI-related risks drawn from 43 taxonomies, organized by a causal taxonomy (entity, intentionality, timing) and a domain taxonomy covering seven risk areas.
Funding
Funding not disclosed
Founders
Product
Problem
Stakeholders such as academics, auditors, policymakers, and AI developers lack a unified, up‑to‑date reference that captures the full spectrum of AI‑related risks, making it difficult to compare, audit, and mitigate those risks consistently.
Solution
The AI Risk Repository offers a publicly accessible, continuously maintained database of 777 identified AI risks drawn from 43 existing taxonomies. Each risk is classified using a high‑level causal taxonomy (entity, intentionality, timing) and a mid‑level domain taxonomy covering seven risk areas and 23 subdomains. Users can filter, search, and export the data via an online interface or downloadable spreadsheets, enabling systematic analysis and coordinated risk auditing. The repository is built on a systematic literature review and expert consultation, ensuring academic rigor and extensibility as new risks emerge. By providing a common, structured frame of reference, the repository supports more coherent research, policy formulation, and risk management across the AI ecosystem.
Target Audience
Primary users are researchers, auditors, policymakers, and AI product teams who need a systematic, evidence‑based inventory of AI risks for analysis, compliance, and mitigation planning.
Features
- Comprehensive catalog of 777 AI risks sourced from 43 distinct taxonomies
- Dual taxonomy structure: causal factors (entity, intentionality, timing) and domain categories (e.g., discrimination, privacy, misinformation, misuse, HCI, socioeconomic, system safety)
- Interactive web interface with advanced filtering, sorting, and export capabilities
- Regularly updated dataset hosted as online spreadsheets for easy modification and contribution
- Structured metadata enabling quantitative analysis, cross‑risk comparison, and integration into audit workflows
- Open‑access design that allows researchers and regulators to cite, extend, and embed the risk data in their own tools