Publicly ISO Available Specification ISO/PAS8800 First edition Road vehicles - Safety and artificial intelligence 2024-12 Vehiculesroutiers-Sécurite et intelligenceartificielle Referencenumber ISO/PAS 8800:2024(en) @ISO2024 IS0/PAS8800:2024(en) COPYRIGHTPROTECTEDDOCUMENT @IS02024 All rights reserved. Unless otherwise specified, or required in the context of its implementation, no part of this publication may bereproduced orutilizedotherwise in anyformorbyanymeans,electronicormechanical, includingphotocopying,orpostingon the internet or an intranet, without prior written permission. Permission can be requested from either ISO at the address below orIso'smemberbodyinthecountryoftherequester. ISo copyrightoffice CP401·Ch.deBlandonnet8 CH-1214 Vernier, Geneva Phone:+41227490111 Email:[email protected] Website:www.iso.org Published in Switzerland @IS02024-All rights reserved ii IS0/PAS8800:2024(en) Contents Page Foreword ..vi Introduction .vii 1 Scope. 1 2 Normative references 1 3 Terms and definitions 22 3.1 General Al-related definitions 3.2 Data-relateddefinitions .7 3.3 Generalsafety-relateddefinitions .9 3.4 Safety:Root cause-,error-andfailure-related definitions .11 3.5 Miscellaneous definitions. 12 4 Abbreviatedterms. 14 5 Requirementsforconformity .15 5.1 Purpose. 15 5.2 General requirements 15 6 Al within the context of road vehicles system safety engineeringand basic concepts 16 6.1 ApplicationoftheIS026262seriesforthedevelopmentofAIsystems. .16 6.2 Interactionswith encompassingsystem-level safetyactivities .17 6.3 Mappingof abstraction layersbetween theIS026262series,IS0/IE22989andthis document. 20 6.4 ExamplearchitectureforanAIsystem 22 6.5 Types ofAI models 23 6.6 AI technologies of a ML model 23 6.7 Error concepts, fault models and causal models .24 6.7.1 Cause-and-effectchain. .24 6.7.2 Rootcauseclasses. 26 6.7.3 Error classification basedon the safety impact .27 7 AI safety management. 28 7.1 Objectives 28 7.2 Prerequisites and supporting information 28 7.3 General requirements. 28 7.4 ReferenceAI safetylifecycle 31 7.5 IterativedevelopmentparadigmsforAIsystems 33 7.6 Workproducts 34 8 Assuranceargumentsfor AI systems .35 8.1 Objectives 35 8.2 Prerequisites and supporting information 35 8.3 Generalrequirements 36 8.4 AIsystem-specificconsiderations inassurancearguments 36 8.5 Structuring assurance arguments for AI systems. .37 8.5.1 Context of the assurance argument .37 8.5.2 Categories of evidence. 38 8.6 Therole ofquantitativetargets and qualitativearguments 39 8.7 Evaluation ofthe assuranceargument 40 8.8 Workproducts. .41 9 Derivation of Al safetyrequirements 41 9.1 Objectives .41 9.2 Prerequisites and supporting information .42 9.3 General requirements .42 9.4 Generalworkflowforderivingsafetyrequirements .43 9.5 Deriving AI safety requirements on supervised machine learning 46 9.5.1 Theneedforrefined AI safetyrequirements 46 @IS0 2024-All rights reserved ili
ISO PAS 8800-2024 Road vehicles - Safety and artificial intelligence