{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-08-24-001",
    "source_story_id": "tmp-lead-aixamine-cross-dimensional-trust",
    "edition_id": "mp-2026-08-24-morning-0046",
    "edition_url": "https://themachinepress.com/edition/2026-08-24",
    "position": 1,
    "story_type": "lead",
    "section": "safety-security",
    "editorial_classification": "editorial",
    "headline": "The Safe Score Hid the Privacy Collapse",
    "slug": "the-safe-score-hid-the-privacy-collapse",
    "dek": "A black-box audit across more than 120 language models found that safety, security, privacy and benign usefulness can move in different directions.",
    "summary": "A black-box audit across more than 120 language models found that safety, security, privacy and benign usefulness can move in different directions.",
    "body_text": "The aiXamine preprint combines 46 tests across nine services so that alignment, adversarial robustness, privacy and over-refusal are measured as related properties rather than separate leaderboards. Across more than 5,000 runs, the authors report three recurring trade-offs: stronger safety enforcement often rejected more benign requests; privacy was nearly orthogonal to the other trust dimensions; and one off-policy distillation setting collapsed robustness from 56.9 to 2.6 on the same base architecture. These are benchmark findings, not a universal ranking of deployed models, but they make a practical point: a high score on one trust axis cannot certify the others.",
    "why_it_matters": "A black-box audit across more than 120 language models found that safety, security, privacy and benign usefulness can move in different directions.",
    "limitations": [
      "These are benchmark findings, not a universal ranking of deployed models, but they make a practical point: a high score on one trust axis cannot certify the others."
    ],
    "importance": 10,
    "canonical_url": "https://themachinepress.com/story/mp-2026-08-24-001/the-safe-score-hid-the-privacy-collapse",
    "json_url": "https://themachinepress.com/story/mp-2026-08-24-001.json",
    "first_published_at": "2026-08-24T09:00:00.000-04:00",
    "modified_at": "2026-08-24T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-08-24-001-001",
        "text": "A black-box audit across more than 120 language models found that safety, security, privacy and benign usefulness can move in different directions.",
        "source_ids": [
          "source-2026-08-24-001"
        ],
        "qualification": "These are benchmark findings, not a universal ranking of deployed models, but they make a practical point: a high score on one trust axis cannot certify the others."
      }
    ],
    "source_ids": [
      "source-2026-08-24-001"
    ],
    "tags": [
      "LLM safety",
      "privacy",
      "red teaming",
      "evaluation"
    ],
    "image_url": "https://themachinepress.com/issues/2026-08-24/lead-aixamine-trust.png",
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-08-24-001",
      "title": "arXiv preprint 2608.20554",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2608.20554",
      "canonical_url": "https://arxiv.org/abs/2608.20554",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-08-20T16:33:35.000-04:00",
      "accessed_at": "2026-08-24T08:23:00.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-08-24-001-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "The Safe Score Hid the Privacy Collapse",
    "publisher": "The Machine Press",
    "published_at": "2026-08-24T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-08-24-001/the-safe-score-hid-the-privacy-collapse"
  }
}