arXiv:2610.34837 · v1

Beyond Behaviour: A Stimulus-Response Safety Evaluation of Vision-Language-Action Models

首次提交 2026-10-26 12:36

版本更新 2026-10-26 12:36

雷达观测 2026-10-29 00:36

摘要

Humans exhibit biological sensitivity to safety-critical stimuli that can trigger defensive action. Across three models and two video benchmarks, we investigate whether similar responses emerge in vision-language-action models by examining their internal representations and action tendencies in response to visible hazards and harmful consequences. Our findings suggest that task execution may remain the default response even when danger is internally registered.

期刊信息：The Science of Physical AI Safety workshop at CoRL 2026