Automotive silicon in the era of AI, functional safety, and cybersecurity
Automotive silicon in the era of AI, functional safety, and cybersecurity
https://www.edn.com/automotive-silicon-in-the-era-of-ai-functional-safety-and-cybersecurity/
Publish Date: 2026-05-15 04:05:00
Source Domain: www.edn.com
Automotive silicon design is entering a phase where functional safety, cybersecurity and artificial intelligence (AI) can no longer be treated as separate concerns. In connected, software-defined vehicles, safety outcomes depend not only on protection against random hardware faults, but also on resilience to malicious interference and software vulnerabilities. As a result, many of the decisions that determine system safety are now made at the silicon architecture level.
When ISO 26262 was first published in 2011, it marked a major step forward in structuring functional safety for automotive electronics. But the vehicles being designed today are fundamentally different. Autonomous driving, electrification, AI-based perception, vehicle-to-everything (V2X) connectivity, and centralized compute architectures were not primary considerations at the time.
The core objective remains unchanged: to avoid hazards to people. However, the way this objective is achieved is now deeply tied to how safety is architected into semiconductor devices.
Functional safety is no longer just a system-level concern; it’s a design-time challenge for ASIC and SoC engineers. For many safety-critical functions, whether ISO 26262 targets can be met depends on decisions made in the earliest stages of silicon architecture.
A growing and converging standards landscape
The industry has responded to new challenges by expanding the safety and security framework. ISO 26262:2018 addresses functional safety in road vehicles, while ISO 21448 (SOTIF) considers hazards arising from insufficient or incorrect system behavior. ISO/PAS 8800:2024 begins to address the safety implications of AI-based systems.
Alongside these, ISO/SAE 21434 introduces…
