VLA 驾驶模型的推理不忠实问题直接关系到自动驾驶安全，做自动驾驶或具身智能的开发者值得关注——你的模型可能声称停车但实际在开。
该研究首次系统评估了视觉-语言-动作（VLA）驾驶模型在推理忠实性方面的表现，分析了 Alpamayo-R1-10B 在 100 个场景中的 300 次推理。结果显示，模型输出的自然语言推理与轨迹存在显著不忠实：整体推理忠实度仅 42.5%，Chain-of-Causation 匹配场景现实不到一半；在三分之一的行人相关场景中遗漏了 94 个行人；轻微视觉扰动导致 97.7% 的轨迹脆弱；推理-动作一致性仅 48.3%，其中 53.3% 的推理一致性低，37.9% 声称停止但模型继续前进。研究从信息论角度形式化了忠实性，定义了实体和动作忠实性验证标准，并提出了四组件安全架构。
Is VLA Reasoning Faithful? Probing Safety of Chain-of-Causation
We present the first systematic study of faithfulness in Vision-Language-Action (VLA) driving models, analyzing 300 Alpamayo-R1-10B inferences across 100 diverse PhysicalAI-AV scenarios. Our main finding is that output natural-language rationales with trajectories may be significantly unfaithful: (i) overall reasoning fidelity is only 42.5%, with Chain-of-Causation matching scene reality less than half the time; (ii) 94 missed pedestrians in one-third of pedestrian-relevant scenes; (iii) 97.7% trajectory fragility under mild visual perturbations; and (iv) only 48.3% mean reasoning-action consistency, with 53.3% of inferences exhibiting low consistency, including 37.9% of stop-claimed cases where the model continues instead. We formalize faithfulness information-theoretically, define entity and action fidelity with verification criteria, and outline a four-component safety architecture aligned with these results.