{"data":{"id":"7f9f6e89-c0af-4e15-b360-be052c24b3a8","title":"Say the Same, Act Differently: Diagnosing and Steering Text-Action Divergence in Reasoning VLAs","summary":"This research shows that visual perturbations can change the physical action of reasoning vision-language-action models while the generated rationale stays word-for-word identical. Using a text-neutral action-subspace (TNAS) diagnostic on Alpamayo (driving) and InstructVLA (manipulation), the authors find that action-sensitive directions are nearly orthogonal to text-loss gradients. In 154 obstacle-annotated driving scenes, TNAS-guided attacks induced a new collision under an identical rationale in 52.6% of scenes under full-image perturbations, compared with 4.55% for unguided action attacks.","solution":"N/A -- no mitigation discussed in source.","labels":["security","research"],"sourceUrl":"https://arxiv.org/abs/2611.04166v1","publishedAt":"2026-11-03T01:57:41.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"info","attackType":["model_evasion"],"issueType":"research","affectedPackages":null,"affectedPackageNames":null,"affectedPackageRefs":null,"affectedVendors":[],"affectedVendorsRaw":["Alpamayo","InstructVLA"],"classifierModel":"claude-haiku-5-5","classifierPromptVersion":"v4","summaryPromptVersion":"v2","headline":null,"headlinePromptVersion":null,"cvssVector":null,"attackVector":null,"attackComplexity":null,"privilegesRequired":null,"userInteraction":null,"exploitMaturity":null,"epssScore":null,"epssCheckedAt":null,"kevDateAdded":null,"advisoryAliases":null,"affectedPackagesSource":null,"affectedPackagesCheckedAt":null,"patchAvailable":null,"disclosureDate":"2026-11-03T01:57:41.000Z","fixReleasedAt":null,"daysToFix":null,"fixCheckedAt":null,"capecIds":null,"crossRefCount":0,"attackSophistication":"advanced","impactType":["safety","integrity"],"aiComponentTargeted":"model","llmSpecific":false,"classifierConfidence":0.95,"researchCategory":"preprint","atlasIds":null}}