Artificial Intelligence is reshaping the automotive industry at a pace that existing safety standards were not originally designed to accommodate. As vehicles become increasingly software-defined and autonomous driving functions grow in complexity, the intersection of machine intelligence and functional safety has become one of the most critical — and most debated — topics across the automotive standards landscape in 2026.
Why AI Is Challenging Functional Safety Frameworks
Traditional standards such as ISO 26262 were developed around deterministic systems: components with predictable behaviour, defined failure modes, and traceable fault responses. Machine learning introduces a fundamentally different paradigm — systems that learn from data, operate probabilistically, and produce outputs that cannot always be fully anticipated or explained through conventional safety analysis methods.
Advances in electrification, autonomous driving, software-defined vehicles, and intelligent applications are introducing new safety challenges with levels of complexity that have no precedent in conventional automotive development. OEMs and suppliers are actively seeking the next wave of guidance to accommodate these changes.
This creates a significant gap between what current standards require and what these new systems demand — a gap that the automotive standards community is working urgently to close.
The Standards Responding to the Challenge
Several frameworks are evolving in parallel to address these developments:
ISO 26262 — Third Edition in Preparation
The safety community is on the cusp of decision-making for the third edition of ISO 26262, which will need to address the realities of software-defined vehicles and intelligent applications. The upcoming edition is expected to include expansions on automated driving, agile development methodologies, and stronger integration with SOTIF (ISO 21448).
ISO 21448 SOTIF — Safety of the Intended Functionality
Where ISO 26262 addresses failures caused by system malfunctions, SOTIF addresses hazards arising from the intended behaviour of a system — precisely the type of risk that machine learning introduces. Most development teams apply both standards within the same ADAS or autonomous programme. As intelligent systems become more prevalent in vehicles, SOTIF is increasingly central to safety arguments.
ISO/PAS 8800 — Safety and Artificial Intelligence
ISO/CD PAS 8800 is among the new standards under development that will complement the ISO 26262 ecosystem. This emerging framework is expected to provide structured guidance on how intelligent components should be assessed, validated, and integrated within safety-critical automotive systems.
ASPICE for Intelligent System Development
As machine learning components are embedded into safety-relevant systems, ASPICE process discipline becomes increasingly important. The traceability, requirements management, and verification rigour that ASPICE demands provide the process foundation that development in safety-critical contexts requires — even where system outputs cannot be fully predicted.
What This Means for Automotive Suppliers
Modern safety programmes are now expected to address not only hardware and software malfunctions, but also SOTIF scenario modelling, attack tree analysis, and combined risk evaluation — areas where traditional methods are proving insufficient.
For suppliers developing embedded software, ADAS components, or any system incorporating machine learning, this convergence creates both a compliance challenge and a competitive opportunity:
Organisations with mature ISO 26262 and SOTIF processes are better positioned to adapt to upcoming requirements. Those with established ASPICE capability have the process infrastructure to manage development traceability. Suppliers already aligned with ISO/SAE 21434 are equipped to address the cybersecurity dimensions that connectivity introduces.
The Integrated Standards Approach
A typical ADAS programme now addresses ISO 26262 for electrical and electronic malfunctions, ISO 21448 for functional insufficiencies in nominal operation, and ASPICE for process capability — all within the same development programme.
This integration is no longer optional. The ability to manage multiple overlapping standards simultaneously — and demonstrate that capability to OEM customers — is becoming a core differentiator in supplier qualification.
Preparing Your Organisation Now
The normative landscape is evolving rapidly, but the direction is clear: standards will become more demanding, OEM expectations will increase, and suppliers without structured processes in place will face growing barriers to programe participation.
Organisations that invest now in building integrated capability across ISO 26262, SOTIF, ASPICE, and ISO/SAE 21434 will be significantly better positioned to navigate the next wave of requirements — and to demonstrate to their customers that their development processes are ready for the intelligent vehicle of tomorrow.
At Ariol Automotive, we help organisations across the automotive supply chain build and integrate this capability — from gap analysis through to full assessment readiness across multiple standards. Contact us today to find out how we can prepare your organisation for the evolving requirements your customers will demand.
