{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-09-08-014",
    "source_story_id": "tmp-story-one-word-robot-actions",
    "edition_id": "mp-2026-09-08-morning-0061",
    "edition_url": "https://themachinepress.com/edition/2026-09-08",
    "position": 14,
    "story_type": "dispatch",
    "section": "benchmarks-evals",
    "editorial_classification": "editorial",
    "headline": "One Word Changed the Robot Task—and Several Models Lost the Constraints",
    "slug": "one-word-changed-the-robot-task-and-several-models-lost-the-constraints",
    "dek": "A real-robot benchmark separates instructions that should preserve an action from those that should change it, then adds multiple constraints.",
    "summary": "A real-robot benchmark separates instructions that should preserve an action from those that should change it, then adds multiple constraints.",
    "body_text": "One Word, Different Action builds paired instructions around physical decision states and executable actions. Some wording changes leave the task intact and test decision invariance; others change the task and test decision sensitivity. Modern models approached saturation when a single constraint changed, according to the authors, but several degraded when one action decision had to integrate multiple requirements. The benchmark suggests the remaining problem is less about noticing an isolated word and more about composing constraints reliably under real visual grounding.",
    "why_it_matters": "A real-robot benchmark separates instructions that should preserve an action from those that should change it, then adds multiple constraints.",
    "limitations": [
      "Modern models approached saturation when a single constraint changed, according to the authors, but several degraded when one action decision had to integrate multiple requirements."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-08-014/one-word-changed-the-robot-task-and-several-models-lost-the-constraints",
    "json_url": "https://themachinepress.com/story/mp-2026-09-08-014.json",
    "first_published_at": "2026-09-08T09:00:00.000-04:00",
    "modified_at": "2026-09-08T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-09-08-014-001",
        "text": "A real-robot benchmark separates instructions that should preserve an action from those that should change it, then adds multiple constraints.",
        "source_ids": [
          "source-2026-09-08-014"
        ],
        "qualification": "Modern models approached saturation when a single constraint changed, according to the authors, but several degraded when one action decision had to integrate multiple requirements."
      }
    ],
    "source_ids": [
      "source-2026-09-08-014"
    ],
    "tags": [
      "embodied reasoning",
      "language grounding",
      "robot benchmark"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-09-08-014",
      "title": "arXiv preprint 2609.05260",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2609.05260",
      "canonical_url": "https://arxiv.org/abs/2609.05260",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-09-04T11:21:58.000-04:00",
      "accessed_at": "2026-09-08T08:20:00.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-09-08-014-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "One Word Changed the Robot Task—and Several Models Lost the Constraints",
    "publisher": "The Machine Press",
    "published_at": "2026-09-08T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-08-014/one-word-changed-the-robot-task-and-several-models-lost-the-constraints"
  }
}