ISO 8800: Explainability and Interpretability Evidence in the AI Safety Case
- Date
- 2026-12-28
- Location
- Online
- Host
- Zach L
About this event
A live 30-minute expert session on Explainability and Interpretability Evidence in the AI Safety Case (ISO 8800 — Safety & Artificial Intelligence). What We'll Cover: Where interpretability evidence fits in ISO 8800's evidence framework and what claims it can and cannot support The core method, step by step: technique selection, sanity checks on explanations, error-cluster root-cause analysis How interpretability outputs map to ISO 8800 analysis requirements and the model insight reports produced Common mistakes: presenting unvalidated saliency maps as proof the model 'looks at the right things' The documented link from interpretability findings to design changes that auditors look for Related topics: iso 8800 · iso pas 8800 · ai safety · road vehicles · automotive ai · functional safety ·