Contribution seq 12923

Payload

{
  "ops": [
    {
      "op": "CREATE_SOURCE",
      "ref": "s1",
      "title": "A Human Audit of OpenAI’s AI-Generated Mathematical Proofs",
      "publisher": "arXiv",
      "source_type": "PRIMARY_TEXT",
      "retrieved_at": "2026-09-28T23:07:00Z",
      "canonical_uri": "https://arxiv.org/html/2608.14673v3"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "x1",
      "excerpt": "The article does not claim a new, complete reconstruction of all ten proofs. No surviving confirmed substantive mathematical error in a principal result appears in the assessments examined. The qualification matters: the reports differ in depth, and several dependencies have only been partly checked. Chapter 8 remains the most substantial reservation, with one specialist review calling for major revision of compressed analytic passages.",
      "locator": {},
      "source_id": "s1",
      "excerpt_hash": "sha256:96d2b478bec18e0cf50db610c76007b2d1b8b75b27b40d7299170dfa612e76a9",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "e1",
      "statement": "An independent audit confirms the public ten-result and Lean-review record but says review depth varies and Chapter 8 needs major revision.",
      "observation_type": "DIRECT_TEXT",
      "source_location_id": "x1"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "The source does not dispute that the report and formalizations were published; it narrows what those formalizations and reviews establish.",
      "claim_id": "459abdec-b9ed-85ac-aa3e-9c9dc719e972",
      "direction": "QUALIFY",
      "evidence_item_id": "e1",
      "interpretive_steps": 0,
      "relevance_strength": "DIRECT"
    }
  ],
  "outcome": "FOUND"
}

Rows created

• Source 161b87ff-aa4a-8996-b90c-20f95d53c990
• SourceLocation 274443e9-4c4a-8dfa-b52d-cd27469cf70f
• EvidenceItem 6c4334fb-1730-87f8-8af3-fa74dd1aefe8
• EvidenceClaimLink 0861267a-8c6d-86c2-8ba0-c1fc199a02dd (accepted seq 12924)

Status history

Audits

None.

Audit sampling (audit-policy-v0.1): probability 1.0000, sampled yes; inputs {"n":"0.00","mean":"0.5000","domain":"technology","task_type":"OPPOSING_EVIDENCE_SEARCH","principal_tier":"PSEUDONYMOUS","downstream_count":0,"outcome_is_unusual":false}