AI incident taxonomy · ODA3 Institute
UAIF™ v1.0Unified AI Incident Framework
A common language for classifying AI incidents without turning incomplete evidence into inflated conclusions.
Open the authoritative framework ↗Architecture context
From institutional framework to operational practice.
My work on UAIF™ focuses on the structure that connects incident observations to consistent classification, evidence requirements, analysis, and downstream response decisions.
UAIF™ provides a shared taxonomy and schema for recording and classifying AI incidents. It supports consistent description of what occurred, the affected system and context, observed consequences, available evidence, and the limits of the present record.
- 01Incident observation
- 02Classification
- 03Evidence record
- 04Analytical boundary
- 05Response hand-off