DEKRA CEO advocates for explainable, resilient, and safe AI systems
“The real challenge is whether people can trust the AI systems operating on our roads”
Artificial intelligence (AI) in vehicles can only deliver on its full potential if people have trust in these systems – not only at launch, but throughout their entire lifecycle. This was the central message emphasized by Stan Zurkiewicz, CEO of the Testing, Inspection, and Certification organization DEKRA, in his address to business and AI technology leaders, developers and experts at the Adopt AI International Summit in Paris, France.
“We are entering a decade in which AI determines the functioning of safety-critical systems, and where 75 percent of innovations in the automotive industry will be software-driven,” said Zurkiewicz. “The real challenge is not whether we can build powerful AI systems. The real challenge is whether we can build them in a way that people can trust the systems operating on our roads.”
The DEKRA CEO warned that a single approval stamp is no longer enough: “As automotive systems become more intelligent and software-driven, traditional one-time validation methods are no longer sufficient,” said Zurkiewicz. Digital Trust requires continuous evaluation, ensuring that digital systems perform correctly, securely, and consistently – even after updates, or evolving operating conditions. Trust must be maintained throughout the entire lifecycle of a vehicle.
Three essential elements of Digital Trust
According to Zurkiewicz, establishing Digital Trust for automotive AI requires the integration of three essential domains:
o Safe and ethical AI, so that it can be understood how decisions are made and whether systems behave correctly in dynamic, real-world conditions – especially as AI-enabled components continuously learn and adapt.
o Cybersecurity, so that the integrity and confidentiality of vehicle systems cannot be compromised across increasingly connected architectures.
o Functional safety, so that potential errors remain predictable, and that a vehicle is still controllable in the event of a malfunction.
As AI, cybersecurity, and functional safety increasingly converge, the safety model of the past century – built for deterministic, mechanical systems – must evolve to one designed for intelligence. This requires combining classical safety engineering with new AI-specific techniques such as model interpretability, strong data governance, and continuous performance monitoring.
Integrated Digital Trust Services
DEKRA has combined the testing and certification of these three domains into its Digital Trust Services. As the world’s largest non-listed organization in the field of testing, inspection and certification, DEKRA supports automotive and digital industry stakeholders by building trust into systems from the very beginning and maintaining it throughout the entire lifecycle.
This includes AI model validation, cybersecurity assurance, functional safety assessment, and support with current and emerging regulatory frameworks such as ISO 21434 (cybersecurity), ISO 26262 (functional safety), ISO 8800 (AI safety), and UNECE R155 (cybersecurity management system).
“Integrating these disciplines results in Digital Trust,” emphasized Zurkiewicz. “The world does not need more hype around AI or new technologies. What it needs is confidence in these systems and components: It needs proof.”
About the Adopt AI Summit
The Adopt AI International Summit (most recently on November 25 and 26, 2025) in Paris, France, is the leading international conference dedicated to accelerating the adoption of AI across multiple sectors. It brings together leaders from business, government, and technology to share visions, insights, and applications related to AI.