AI Risk Repository

Field Explanations

Causal Taxonomy: Classifies risks based on their causal factors such as the timing and intent.

Domain Taxonomy: Classifies risks into specific AI risk domains and sub-domains.

Key Fields:

  • Category level: Paper, risk, subcategory of risk, or additional evidence.
  • Risk category & subcategory: Extracted text for risk categories and subcategories.
  • Description: Description of risk or risk category.
  • Entity: Whether the risk is due to AI system, Human, or other.
  • Intent: Whether the risk is due to expected or unexpected outcomes.
  • Timing: Whether the risk occurs pre- or post-deployment.
  • Domain: One of 7 AI risk domains.
  • Sub-domain: One of 23 AI risk sub-domains.
Data from MIT AI Risk Database and website created by Frederic Rohrer.