Technical articles · 2025.10.14
ENFLI: a proposal for fleet-level risk acceptance
Safetronic 2025: Preview – Was ist ein akzeptables Risiko? Ein Vorschlag
Overview
Susanne Ebel and Bernd Müller propose ENFLI by distinguishing a low per-vehicle event rate from accumulated fleet risk. They outline criticality-normalized safety-goal violations (CNSGV), the need to justify expert judgment, and a possible role in defining AI safety qualities and acceptance criteria within their developing SafeAI framework.
Publication and version information
The source records posting on 2025-10-14 and modification on 2026-08-12; the update date is not an exact revision identifier. This is the authors' proposal, not an ISO requirement or a validated guarantee of no incidents. ASIL-based normalization is one possible option, and the method may be unusable without reasonable expert judgment. Its stated mandatory role belongs to the developing SafeAI framework, not to ISO generally. The delivered presentation was not obtained.