{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-10-19-026",
    "source_story_id": "tmp-story-hardeval-security-remediation",
    "edition_id": "mp-2026-10-19-morning-0101",
    "edition_url": "https://themachinepress.com/edition/2026-10-19",
    "position": 15,
    "story_type": "dispatch",
    "section": "safety-security",
    "editorial_classification": "editorial",
    "headline": "The Security Script Looks Plausible Until the Machine Checks It",
    "slug": "the-security-script-looks-plausible-until-the-machine-checks-it",
    "dek": "HardEval executes model-written remediations on live virtual machines and grades the persistent system state with hidden compliance checks.",
    "summary": "HardEval executes model-written remediations on live virtual machines and grades the persistent system state with hidden compliance checks.",
    "body_text": "Security hardening is not complete when a shell script looks reasonable. HardEval turns Red Hat Enterprise Linux 8 STIG controls into execution-verified tasks. A model receives the required outcome but not the implementation, writes a Bash script, and runs it on a live virtual machine; hidden OpenSCAP and OVAL checks determine whether the system actually moved from fail to pass. Across fourteen frontier and open-weight configurations, the authors found large and uneven differences among security domains. A recurring failure was convincing code that did not satisfy the exact compliance condition. The benchmark covers a specific operating system and control set, but it measures the durable state change that code-only evaluations miss.",
    "why_it_matters": "HardEval executes model-written remediations on live virtual machines and grades the persistent system state with hidden compliance checks.",
    "limitations": [
      "Security hardening is not complete when a shell script looks reasonable.",
      "A model receives the required outcome but not the implementation, writes a Bash script, and runs it on a live virtual machine; hidden OpenSCAP and OVAL checks determine whether the system actually moved from fail to pass.",
      "A recurring failure was convincing code that did not satisfy the exact compliance condition."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-19-026/the-security-script-looks-plausible-until-the-machine-checks-it",
    "json_url": "https://themachinepress.com/story/mp-2026-10-19-026.json",
    "first_published_at": "2026-10-19T09:00:00.000-04:00",
    "modified_at": "2026-10-19T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-10-19-026-001",
        "text": "HardEval executes model-written remediations on live virtual machines and grades the persistent system state with hidden compliance checks.",
        "source_ids": [
          "source-2026-10-19-015"
        ],
        "qualification": "Security hardening is not complete when a shell script looks reasonable."
      }
    ],
    "source_ids": [
      "source-2026-10-19-015"
    ],
    "tags": [
      "security hardening",
      "agent evaluation",
      "compliance"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-10-19-015",
      "title": "arXiv preprint 2610.23685, original v1",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2610.23685v1",
      "canonical_url": "https://arxiv.org/abs/2610.23685v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-10-16T12:14:58.000-04:00",
      "accessed_at": "2026-10-19T08:27:48.782-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-10-19-026-001"
      ]
    }
  ],
  "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 Security Script Looks Plausible Until the Machine Checks It",
    "publisher": "The Machine Press",
    "published_at": "2026-10-19T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-19-026/the-security-script-looks-plausible-until-the-machine-checks-it"
  }
}