研究显示 Reasoning VLA 的文字解释可与实际动作完全脱钩
Say the Same, Act Differently: Diagnosing and Steering Text-Action Divergence in Reasoning VLAs
这篇论文发现视觉扰动能让驾驶模型撞车，但生成的安全解释一字不改，靠读模型解释做监督的人得看看。
arXiv 论文提出 TNAS 诊断方法，测试 Alpamayo 与 InstructVLA 两个 Reasoning VLA 模型，发现视觉扰动可以让物理动作发生改变而文字解释保持逐字相同。在 Alpamayo 的 154 个带障碍物标注的驾驶场景中，TNAS 引导的攻击在全图扰动下有 52.6% 的场景在解释不变的情况下引发新碰撞，前视摄像头局部贴片下为 35.1%。相同隐状态扰动幅度下，沿该方向的干预对动作的影响是文本梯度扰动的 4.64 到 47.8 倍。论文结论是文字解释不变不能证明动作安全，监控系统必须直接校验预测的动作通道。
Say the Same, Act Differently: Diagnosing and Steering Text-Action Divergence in Reasoning VLAs
Reasoning vision-language-action (VLA) models pair physical control with natural-language rationales that supervisors can read to monitor embodied safety. Prior attacks on VLA reasoning corrupt the rationale itself and leave visible evidence in the text. We show that visual perturbations can instead alter the physical action while keeping the generated rationale word-for-word identical. At the shared text/action interface of Alpamayo for driving and InstructVLA for manipulation, our text-neutral action-subspace (TNAS) diagnostic shows that the locally most action-sensitive direction is nearly orthogonal to the active text-loss gradients, with 1.87% and 5.28% norm overlap. At equal hidden-state perturbation norms across both models, interventions along this text-neutral residual shift actions 4.64x to 47.8x more than text-gradient perturbations, whereas text-gradient perturbations induce 21.2x to 277x more text drift. Steering visual perturbations along this direction reaches the same decoupling from the image. Across 154 obstacle-annotated driving scenes in Alpamayo, TNAS-guided attacks induce a new collision under a word-for-word identical rationale in 52.6% of scenes under full-image perturbations and 35.1% under a localized forward-camera patch, compared with 4.55% for unguided action attacks. Relative to a text-preserving baseline, TNAS guidance raises exact rationale preservation from 83.0% to 89.5% and cuts teacher-forced text drift by 68.4% at a comparable collision rate. In InstructVLA, TNAS guidance cuts text drift by 40.5% and preserves the exact response on 85.0% of examples alongside a 1.73-unit action-chunk shift. An unchanged rationale therefore cannot certify a safe action, so monitors for reasoning VLAs must verify the predicted action channel directly.