SHAP Fails Autonomous Driving Safety Standards — Causal XAI Required

A rubric derived from AMLAS, ISO 26262, ISO 21448, and ISO/PAS 8800 scores six XAI method classes against 19 testable criteria across 7 lifecycle stages. SHAP, the most-used ADS explainability method, cannot structurally satisfy the directed cause-and-effect evidence requirements these standards demand. Causal XAI closes rubric gaps of 62% at hazard identification and 50% at incident investigation.