Keysight and University of York Partner for SDV AI Safety Assurance

Keysight Technologies and the Centre for Assuring Autonomy (CfAA) at the University of York have announced a research collaboration focused on validating artificial intelligence (AI) systems within software-defined vehicles (SDVs). The initiative aims to develop structured engineering methodologies and auditable safety cases to demonstrate AI safety and reliability in compliance with emerging automotive standards.

Context of the Cooperation
Artificial intelligence is becoming foundational across advanced driver assistance systems (ADAS), autonomous driving functions, and modern software-defined vehicles. However, automotive original equipment manufacturers (OEMs) and Tier-1 suppliers face increasing pressure to prove that non-deterministic AI models operate reliably, safely, and as intended throughout the complete vehicle lifecycle. Providing verifiable, auditable proof under evolving safety regulations remains a major industry bottleneck that can introduce significant development risks and delay vehicle rollouts.

To address these compliance and testing challenges, cooperation was established between electronic test and measurement provider Keysight Technologies and the University of York’s Centre for Assuring Autonomy. The partnership combines the university’s academic safety engineering research with Keysight’s commercial AI validation techniques to bridge the gap between academic AI safety principles and practical automotive engineering.

Technical Solution and Responsibilities
The technological framework focuses on creating evidence-driven validation workflows to support conformance with ISO/PAS 8800, the emerging international standard for artificial intelligence in road vehicles. Responsibilities are allocated across the academic and industrial partners:

At a system level, the research explores measurable, academically grounded safety-scoring methodologies and practical frameworks for generating auditable AI safety evidence. This structured approach enables engineering teams to build defensible AI safety arguments and evaluate machine learning models throughout the vehicle lifecycle. Simon Burton, Chair in Systems Safety at the University of York, stated that ensuring increasingly integral AI systems can be evaluated using robust, evidence-based approaches is essential for the automotive industry. Lukas Klose, Head of the Automotive AI Solution Center at Keysight, noted that the initiative aims to develop methodologies that help engineering teams generate structured evidence for AI safety cases and deploy trustworthy AI technologies in conformance with ISO/PAS 8800.

Deployment or Implementation
The implementation strategy focuses on translating theoretical safety assurance models into production-ready software tools for automakers and suppliers.

The research findings will directly inform the future development of Keysight’s AI Software Integrity Builder, expanding its automated capabilities for generating safety arguments and validation evidence. By providing standardized, auditable validation workflows aligned with automotive industry expectations, the resulting methodologies are designed to reduce development risk and accelerate the market deployment of AI-enabled automotive platforms.

Edited by Romila DSilva, Induportals Editor, with AI assistance.