{
  "tool_id": "art-315-ab2013-training-data-disclosure-linter",
  "note": "golden_hash filled by `node golden-parity.test.mjs --update`.",
  "vectors": [
    {
      "name": "empty-insufficient",
      "description": "No disclosure supplied; all 12 datapoints missing, finite-gate.",
      "policy_parameters": {},
      "output_payload": {
        "per_datapoint": [
          {
            "datapoint": "dataset_sources_or_owners",
            "status": "missing"
          },
          {
            "datapoint": "dataset_purpose_alignment",
            "status": "missing"
          },
          {
            "datapoint": "number_of_datapoints",
            "status": "missing"
          },
          {
            "datapoint": "types_of_datapoints",
            "status": "missing"
          },
          {
            "datapoint": "ip_status_copyright_trademark_patent_or_public_domain",
            "status": "missing"
          },
          {
            "datapoint": "purchased_or_licensed",
            "status": "missing"
          },
          {
            "datapoint": "includes_personal_information",
            "status": "missing"
          },
          {
            "datapoint": "includes_aggregate_consumer_information",
            "status": "missing"
          },
          {
            "datapoint": "cleaning_processing_or_modification_description",
            "status": "missing"
          },
          {
            "datapoint": "synthetic_data_use",
            "status": "missing"
          },
          {
            "datapoint": "collection_time_period",
            "status": "missing"
          },
          {
            "datapoint": "collection_dates_or_version",
            "status": "missing"
          }
        ],
        "missing_datapoints": [
          "dataset_sources_or_owners",
          "dataset_purpose_alignment",
          "number_of_datapoints",
          "types_of_datapoints",
          "ip_status_copyright_trademark_patent_or_public_domain",
          "purchased_or_licensed",
          "includes_personal_information",
          "includes_aggregate_consumer_information",
          "cleaning_processing_or_modification_description",
          "synthetic_data_use",
          "collection_time_period",
          "collection_dates_or_version"
        ],
        "present_count": 0,
        "total_datapoints": 12,
        "all_present": false,
        "insufficient_evidence": true,
        "statute_citation": "Cal. Bus. & Prof. Code §22757.7 (AB 2013), eff. 2026-01-01 -- DRAFT-PINNED against secondary-source summaries"
      },
      "golden_hash": "6380fc67165fc86a66cc2b75610f67a05cea6cad2eebb9abf7ce835964cb7b5f"
    },
    {
      "name": "all-12-present",
      "description": "Every AB 2013 datapoint present in the supplied disclosure.",
      "policy_parameters": {
        "disclosure": {
          "dataset_sources_or_owners": "Common Crawl, licensed news corpus",
          "dataset_purpose_alignment": "General-purpose language modeling",
          "number_of_datapoints": "~50B tokens (estimated)",
          "types_of_datapoints": "Web text, labeled QA pairs",
          "ip_status_copyright_trademark_patent_or_public_domain": "Mixed; some copyrighted excerpts under fair-use assertion",
          "purchased_or_licensed": "Partially licensed (news corpus)",
          "includes_personal_information": "No",
          "includes_aggregate_consumer_information": "No",
          "cleaning_processing_or_modification_description": "Deduplicated, PII-scrubbed",
          "synthetic_data_use": "Yes, for RLHF preference pairs",
          "collection_time_period": "2019-2025",
          "collection_dates_or_version": "v3, collected through 2025-12-01"
        }
      },
      "output_payload": {
        "per_datapoint": [
          {
            "datapoint": "dataset_sources_or_owners",
            "status": "present"
          },
          {
            "datapoint": "dataset_purpose_alignment",
            "status": "present"
          },
          {
            "datapoint": "number_of_datapoints",
            "status": "present"
          },
          {
            "datapoint": "types_of_datapoints",
            "status": "present"
          },
          {
            "datapoint": "ip_status_copyright_trademark_patent_or_public_domain",
            "status": "present"
          },
          {
            "datapoint": "purchased_or_licensed",
            "status": "present"
          },
          {
            "datapoint": "includes_personal_information",
            "status": "present"
          },
          {
            "datapoint": "includes_aggregate_consumer_information",
            "status": "present"
          },
          {
            "datapoint": "cleaning_processing_or_modification_description",
            "status": "present"
          },
          {
            "datapoint": "synthetic_data_use",
            "status": "present"
          },
          {
            "datapoint": "collection_time_period",
            "status": "present"
          },
          {
            "datapoint": "collection_dates_or_version",
            "status": "present"
          }
        ],
        "missing_datapoints": [],
        "present_count": 12,
        "total_datapoints": 12,
        "all_present": true,
        "insufficient_evidence": false,
        "statute_citation": "Cal. Bus. & Prof. Code §22757.7 (AB 2013), eff. 2026-01-01 -- DRAFT-PINNED against secondary-source summaries"
      },
      "golden_hash": "2f2eaf5ffe660007391306c0efae2252f8c010d19f0a45f5031e64b0c1b57822"
    },
    {
      "name": "two-datapoints-missing",
      "description": "personal-information and synthetic-data-use disclosures omitted.",
      "policy_parameters": {
        "disclosure": {
          "dataset_sources_or_owners": "Common Crawl, licensed news corpus",
          "dataset_purpose_alignment": "General-purpose language modeling",
          "number_of_datapoints": "~50B tokens (estimated)",
          "types_of_datapoints": "Web text, labeled QA pairs",
          "ip_status_copyright_trademark_patent_or_public_domain": "Mixed; some copyrighted excerpts under fair-use assertion",
          "purchased_or_licensed": "Partially licensed (news corpus)",
          "includes_aggregate_consumer_information": "No",
          "cleaning_processing_or_modification_description": "Deduplicated, PII-scrubbed",
          "collection_time_period": "2019-2025",
          "collection_dates_or_version": "v3, collected through 2025-12-01"
        }
      },
      "output_payload": {
        "per_datapoint": [
          {
            "datapoint": "dataset_sources_or_owners",
            "status": "present"
          },
          {
            "datapoint": "dataset_purpose_alignment",
            "status": "present"
          },
          {
            "datapoint": "number_of_datapoints",
            "status": "present"
          },
          {
            "datapoint": "types_of_datapoints",
            "status": "present"
          },
          {
            "datapoint": "ip_status_copyright_trademark_patent_or_public_domain",
            "status": "present"
          },
          {
            "datapoint": "purchased_or_licensed",
            "status": "present"
          },
          {
            "datapoint": "includes_personal_information",
            "status": "missing"
          },
          {
            "datapoint": "includes_aggregate_consumer_information",
            "status": "present"
          },
          {
            "datapoint": "cleaning_processing_or_modification_description",
            "status": "present"
          },
          {
            "datapoint": "synthetic_data_use",
            "status": "missing"
          },
          {
            "datapoint": "collection_time_period",
            "status": "present"
          },
          {
            "datapoint": "collection_dates_or_version",
            "status": "present"
          }
        ],
        "missing_datapoints": [
          "includes_personal_information",
          "synthetic_data_use"
        ],
        "present_count": 10,
        "total_datapoints": 12,
        "all_present": false,
        "insufficient_evidence": false,
        "statute_citation": "Cal. Bus. & Prof. Code §22757.7 (AB 2013), eff. 2026-01-01 -- DRAFT-PINNED against secondary-source summaries"
      },
      "golden_hash": "b2b3f8a6159ceb75ec955188d52bba11ec1116bb9e4d9f365f5f94ec04434b15"
    }
  ]
}