runtime-guardrails

Tag

Cards List
#runtime-guardrails

Silent Failures in Physical AI: A Literature Review of Runtime Action Authorization for Autonomous Systems

Hugging Face Daily Papers · 2026-05-23 Cached

This literature review identifies and analyzes the problem of silent failures in physical AI systems, where black-box models may execute harmful actions without detection. It proposes a taxonomy of runtime guardrail functions and outlines evaluation requirements for safe autonomous systems.

0 favorites 0 likes
← Back to home

Submit Feedback