"The safety failures we are not instrumenting": Hidden safety-critical challenges in modern AI systems

A position paper argues that current AI safety discussions focus on visible output anomalies but overlook hidden systemic failures arising from interactions across the entire deployment stack. The authors propose four common characteristics: trustworthiness, distributionality, temporal extension, and correction degradation, and build a five-layer diagnostic framework of cognitive integrity, control integrity, temporal integrity, organizational integrity, and ecological integrity to characterize issues such as calibration debt, uncertainty laundering, instruction authority collapse, action amplification, safety drift, memory contamination, and synthetic evidence pollution.