{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-08-30-012",
    "source_story_id": "tmp-story-multi2av-safety",
    "edition_id": "mp-2026-08-30-morning-0052",
    "edition_url": "https://themachinepress.com/edition/2026-08-30",
    "position": 12,
    "story_type": "dispatch",
    "section": "safety",
    "editorial_classification": "editorial",
    "headline": "Benign Inputs Combined Into Harm",
    "slug": "benign-inputs-combined-into-harm",
    "dek": "Multi2AV-Safety tests all 11 multi-input combinations of text, image, audio and video conditioning.",
    "summary": "Multi2AV-Safety tests all 11 multi-input combinations of text, image, audio and video conditioning.",
    "body_text": "The 11,024-instance benchmark is designed around harm that appears only when modalities interact, as well as explicit harmful cues diluted by benign context. Representative safety guards missed both kinds of compositional evidence across time and modality in the authors' evaluation. The dataset is scheduled for release in October 2026, so current claims rest on the paper's reported protocol rather than an independently inspectable public benchmark.",
    "why_it_matters": "Multi2AV-Safety tests all 11 multi-input combinations of text, image, audio and video conditioning.",
    "limitations": [
      "The 11,024-instance benchmark is designed around harm that appears only when modalities interact, as well as explicit harmful cues diluted by benign context."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-08-30-012/benign-inputs-combined-into-harm",
    "json_url": "https://themachinepress.com/story/mp-2026-08-30-012.json",
    "first_published_at": "2026-08-30T09:00:00.000-04:00",
    "modified_at": "2026-08-30T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-08-30-012-001",
        "text": "Multi2AV-Safety tests all 11 multi-input combinations of text, image, audio and video conditioning.",
        "source_ids": [
          "source-2026-08-30-012"
        ],
        "qualification": "The 11,024-instance benchmark is designed around harm that appears only when modalities interact, as well as explicit harmful cues diluted by benign context."
      }
    ],
    "source_ids": [
      "source-2026-08-30-012"
    ],
    "tags": [
      "multimodal safety",
      "audio-video generation",
      "benchmarks"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-08-30-012",
      "title": "arXiv preprint 2608.26535",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2608.26535",
      "canonical_url": "https://arxiv.org/abs/2608.26535",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-08-26T22:12:25.000-04:00",
      "accessed_at": "2026-08-30T08:23:14.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-08-30-012-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "Benign Inputs Combined Into Harm",
    "publisher": "The Machine Press",
    "published_at": "2026-08-30T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-08-30-012/benign-inputs-combined-into-harm"
  }
}