Why LLMs Are Fundamentally Problematic in Safety-Critical Systems

Large Language Models are powerful tools for text generation, but they introduce structural risks when used in safety-critical or regulated environments.

Large Language Models are powerful tools for text generation, but they introduce structural risks when used in safety-critical or regulated environments under ISO 26262 and SOTIF.

Key issues

Non-deterministic behavior
Lack of formal guarantees
Hallucinations without verifiable causal chains
No direct path to certification under ISO 26262 or ISO 21448

What regulated domains require

In domains such as automotive, aerospace, and defense, systems must be deterministic, verifiable, traceable, and certifiable. This is why purely statistical AI approaches struggle in these contexts.

Alternatives exist

Model-based, symbolic, and declarative AI approaches can transform natural language into formal representations that allow reasoning, verification, and compliance with safety standards.

The future of AI in regulated industries is not about replacing engineering rigor. It is about restoring it.

Schedule an intro call All writing

Functional Safety/Sep 18, 2026

Runtime Monitoring for SOTIF: A Monitor Ensemble for AI-Based Perception

A new peer-reviewed, open-access paper in Discover Artificial Intelligence (Springer Nature) sets out a runtime monitor ensemble that supervises AI-based perception during operation, giving a defensible basis for the safe release and series operation of autonomous driving functions under SOTIF.

Computer Vision/Jul 15, 2026

The Affine Ceiling: Sharp on Affine Motion, Blind at Depth

Formal motion estimation is dismissed as good only for small motion. A benchmark against pixel-exact ground truth shows that is false: large scaling and rotation are solved to sub-pixel accuracy. The real ceiling is the affine assumption, and it collapses at exactly the depth boundary that 3D reconstruction is made of.

Computer Vision/Jul 10, 2026

Your Perception Stack Is a Snapshot Machine

Most perception stacks reason one frame at a time. The network detects, tracking is bolted on afterwards, and the system never really carries the world forward. A snapshot machine cannot validate cleanly, because the thing that would make its output trustworthy is the thing it discards between frames: continuity.