{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-10-11-005",
    "source_story_id": "tmp-story-alignment-value-representations",
    "edition_id": "mp-2026-10-11-morning-0093",
    "edition_url": "https://themachinepress.com/edition/2026-10-11",
    "position": 5,
    "story_type": "dispatch",
    "section": "safety-security",
    "editorial_classification": "editorial",
    "headline": "A Value's Name Predicted Less Than Its Activation",
    "slug": "a-value-s-name-predicted-less-than-its-activation",
    "dek": "Model activations forecast how post-training on one value affected unseen values better than text descriptions did.",
    "summary": "Model activations forecast how post-training on one value affected unseen values better than text descriptions did.",
    "body_text": "Researchers define alignment generalization prediction as forecasting how tuning for one value changes behavior across held-out values. Their analysis covers 66 values drawn from modern alignment targets. Representations built from model activations while values were applied in context reached a reported correlation of 0.45 with the measured generalization matrix, versus 0.05 for description-based baselines. The authors also connect similarity within a multi-value target to robustness and report early evidence for a shared value space across models. The results do not make value alignment predictable in general; they show that internal, contextual representations carried more information than labels in the tested setup.",
    "why_it_matters": "Model activations forecast how post-training on one value affected unseen values better than text descriptions did.",
    "limitations": [
      "The results do not make value alignment predictable in general; they show that internal, contextual representations carried more information than labels in the tested setup."
    ],
    "importance": 9,
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-11-005/a-value-s-name-predicted-less-than-its-activation",
    "json_url": "https://themachinepress.com/story/mp-2026-10-11-005.json",
    "first_published_at": "2026-10-11T09:00:00.000-04:00",
    "modified_at": "2026-10-11T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-10-11-005-001",
        "text": "Model activations forecast how post-training on one value affected unseen values better than text descriptions did.",
        "source_ids": [
          "source-2026-10-11-005"
        ],
        "qualification": "The results do not make value alignment predictable in general; they show that internal, contextual representations carried more information than labels in the tested setup."
      }
    ],
    "source_ids": [
      "source-2026-10-11-005"
    ],
    "tags": [
      "alignment",
      "value representations",
      "generalization"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-10-11-005",
      "title": "arXiv preprint 2610.12410, original v1",
      "publisher": "arXiv / submitting research authors",
      "url": "https://arxiv.org/abs/2610.12410v1",
      "canonical_url": "https://arxiv.org/abs/2610.12410v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-10-08T13:47:26.000-04:00",
      "accessed_at": "2026-10-11T08:21:27.764-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-10-11-005-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "A Value's Name Predicted Less Than Its Activation",
    "publisher": "The Machine Press",
    "published_at": "2026-10-11T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-11-005/a-value-s-name-predicted-less-than-its-activation"
  }
}