{
  "schema": "ROOT0-ORG-GIT-AUDIT-v1.0",
  "snapshot_date": "2026-09-14",
  "historical_cutoff": "2025-12-31",
  "scope": {
    "sources": [
      "GitHub public API",
      "Anthropic first-party policy/research pages",
      "OpenAI first-party policy/safety pages"
    ],
    "search_observation": {
      "anthropics_query": "organization=anthropics, query=anthropics",
      "anthropics_visible_matches": 15,
      "openai_query": "organization=openai, query=openai",
      "openai_visible_matches": 73,
      "caveat": "Search-result observations, not complete organization inventories or moral scores."
    },
    "axes": [
      "safety/existential-risk governance",
      "operational controls",
      "developer/commercial surface",
      "provenance/reproducibility"
    ]
  },
  "corpus_ingest": {
    "status": "blocked",
    "sha256": null,
    "reason": "The GitHub connector can write UTF-8 text but this session has no byte-stream access to attachment copies or local hashing interface.",
    "required_fields": [
      "filename",
      "mime",
      "byte_length",
      "sha256",
      "format_metadata",
      "source_derivative_relation",
      "cutoff_classification",
      "signed_commit_or_release"
    ],
    "github_sha_note": "GitHub object SHAs are not SHA-256 file digests and must not be substituted."
  },
  "anthropic": {
    "first_party_safety_sources": [
      {
        "title": "Constitutional AI: Harmlessness from AI Feedback",
        "url": "https://www.anthropic.com/research/constitutional-ai-harmlessness-from-ai-feedback",
        "date": "2022-12-15",
        "cutoff_classification": "admissible"
      },
      {
        "title": "Claude's Constitution",
        "url": "https://www.anthropic.com/constitution",
        "date": null,
        "cutoff_classification": "live_observation"
      },
      {
        "title": "Responsible Scaling Policy",
        "url": "https://www.anthropic.com/responsible-scaling-policy",
        "versions_on_or_before_cutoff": [
          {
            "version": "1.0",
            "effective": "2023-09-19"
          },
          {
            "version": "2.0",
            "effective": "2024-10-15"
          },
          {
            "version": "2.1",
            "effective": "2025-03-31"
          },
          {
            "version": "2.2",
            "effective": "2025-05-14"
          }
        ],
        "later_versions": "live_observation",
        "cutoff_classification": "mixed"
      },
      {
        "title": "Anthropic research index",
        "url": "https://www.anthropic.com/research",
        "date": null,
        "cutoff_classification": "live_observation"
      }
    ],
    "representative_repositories": [
      {
        "repo": "anthropics/toy-models-of-superposition",
        "origin": "2022-09-14",
        "license": "MIT",
        "signal": [
          "research",
          "interpretability"
        ],
        "cutoff": "admissible",
        "url": "https://github.com/anthropics/toy-models-of-superposition"
      },
      {
        "repo": "anthropics/anthropic-sdk-python",
        "origin": "2023-01-17",
        "license": "MIT",
        "signal": [
          "API",
          "developer"
        ],
        "controls": [
          "LICENSE",
          "CONTRIBUTING.md",
          "SECURITY.md",
          ".github"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/anthropics/anthropic-sdk-python"
      },
      {
        "repo": "anthropics/anthropic-sdk-typescript",
        "origin": "2023-01-30",
        "license": "MIT",
        "signal": [
          "API",
          "developer"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/anthropics/anthropic-sdk-typescript"
      },
      {
        "repo": "anthropics/political-neutrality-eval",
        "origin": "2025-11-12",
        "tip_date": "2025-11-13",
        "license": "CC-BY-4.0",
        "signal": [
          "evaluation",
          "political_even_handedness"
        ],
        "cutoff": "admissible",
        "url": "https://github.com/anthropics/political-neutrality-eval"
      },
      {
        "repo": "anthropics/claude-code-base-action",
        "origin": "2025-05-19",
        "license": "MIT",
        "signal": [
          "automation",
          "developer"
        ],
        "controls": [
          "README trust-boundary warning"
        ],
        "cutoff": "admissible",
        "url": "https://github.com/anthropics/claude-code-base-action"
      },
      {
        "repo": "anthropics/claude-plugins-official",
        "origin": "2025-11-20",
        "license": "Apache-2.0",
        "signal": [
          "plugins",
          "ecosystem"
        ],
        "controls": [
          ".github",
          "README third-party trust warning"
        ],
        "cutoff": "admissible",
        "url": "https://github.com/anthropics/claude-plugins-official"
      },
      {
        "repo": "anthropics/courses",
        "origin": "2024-05-30",
        "license": "NOASSERTION",
        "signal": [
          "education",
          "prompt_evals",
          "tool_use",
          "API_cost_guidance"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/anthropics/courses"
      },
      {
        "repo": "anthropics/devcontainer-features",
        "origin": "2025-03-03",
        "tip_date": "2025-12-16",
        "license": "MIT",
        "signal": [
          "developer_distribution"
        ],
        "cutoff": "admissible",
        "url": "https://github.com/anthropics/devcontainer-features"
      }
    ],
    "integrity_samples": [
      {
        "repo": "anthropics/political-neutrality-eval",
        "commit": "c5ed67908b56edc0781f47821241ca44114bd4ff",
        "date": "2025-11-13",
        "verified": true,
        "reason": "valid",
        "url": "https://github.com/anthropics/political-neutrality-eval/commit/c5ed67908b56edc0781f47821241ca44114bd4ff"
      },
      {
        "repo": "anthropics/anthropic-sdk-python",
        "commit": "eb21a4352015686c30f5759e8c2f02d70f5371e2",
        "date": "2026-09-10",
        "verified": true,
        "reason": "valid",
        "url": "https://github.com/anthropics/anthropic-sdk-python/commit/eb21a4352015686c30f5759e8c2f02d70f5371e2"
      }
    ]
  },
  "openai": {
    "first_party_safety_sources": [
      {
        "title": "OpenAI Charter",
        "url": "https://openai.com/charter/",
        "date": "2018-04-09",
        "cutoff_classification": "admissible"
      },
      {
        "title": "Our updated Preparedness Framework",
        "url": "https://openai.com/index/updating-our-preparedness-framework/",
        "date": "2025-04-15",
        "cutoff_classification": "admissible"
      },
      {
        "title": "Safety & responsibility",
        "url": "https://openai.com/safety/",
        "date": null,
        "cutoff_classification": "live_observation"
      },
      {
        "title": "Model Spec repository",
        "url": "https://github.com/openai/model_spec",
        "date": "2025-02-06",
        "cutoff_classification": "admissible_origin"
      }
    ],
    "representative_repositories": [
      {
        "repo": "openai/openai-python",
        "origin": "2020-10-25",
        "license": "Apache-2.0",
        "signal": [
          "API",
          "developer"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/openai/openai-python"
      },
      {
        "repo": "openai/openai-cookbook",
        "origin": "2022-03-11",
        "license": "MIT",
        "signal": [
          "API_examples",
          "developer"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/openai/openai-cookbook"
      },
      {
        "repo": "openai/swarm",
        "origin": "2024-02-22",
        "license": "MIT",
        "signal": [
          "experimental_multi_agent",
          "developer"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/openai/swarm"
      },
      {
        "repo": "openai/model_spec",
        "origin": "2025-02-06",
        "license": "CC0-1.0",
        "signal": [
          "behavior_specification",
          "governance"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/openai/model_spec"
      },
      {
        "repo": "openai/frontier-evals",
        "origin": "2025-03-28",
        "license": "MIT",
        "signal": [
          "frontier_evals",
          "risk_evaluation"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/openai/frontier-evals"
      },
      {
        "repo": "openai/openai-agents-python",
        "origin": "2025-03-11",
        "license": "MIT",
        "signal": [
          "agents",
          "guardrails",
          "handoffs",
          "developer"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/openai/openai-agents-python"
      },
      {
        "repo": "openai/openai-guardrails-python",
        "origin": "2025-04-09",
        "license": "MIT",
        "signal": [
          "safety",
          "compliance"
        ],
        "controls": [
          "LICENSE",
          "CONTRIBUTING.md",
          "SECURITY.md",
          ".github"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/openai/openai-guardrails-python"
      },
      {
        "repo": "openai/gpt-oss",
        "origin": "2025-06-23",
        "license": "Apache-2.0",
        "signal": [
          "open_weights",
          "developer_distribution"
        ],
        "cutoff": "admissible_origin",
        "url": "https://github.com/openai/gpt-oss"
      },
      {
        "repo": "openai/model_spec_evals",
        "origin": "2026-03-24",
        "license": "MIT",
        "signal": [
          "model_spec_compliance_eval"
        ],
        "cutoff": "live_observation",
        "url": "https://github.com/openai/model_spec_evals"
      },
      {
        "repo": "openai/privacy-filter",
        "origin": "2026-04-17",
        "license": "Apache-2.0",
        "signal": [
          "PII_protection",
          "on_prem"
        ],
        "controls": [
          "SECURITY.md"
        ],
        "cutoff": "live_observation",
        "url": "https://github.com/openai/privacy-filter"
      },
      {
        "repo": "openai/codex-security",
        "origin": "2026-07-13",
        "license": "Apache-2.0",
        "signal": [
          "security_analysis",
          "remediation"
        ],
        "cutoff": "live_observation",
        "url": "https://github.com/openai/codex-security"
      }
    ],
    "integrity_samples": [
      {
        "repo": "openai/openai-python",
        "commit": "e12b81d3bbf644ec7045e152d69bc4b68d69cd48",
        "date": "2026-09-11",
        "verified": true,
        "reason": "valid",
        "url": "https://github.com/openai/openai-python/commit/e12b81d3bbf644ec7045e152d69bc4b68d69cd48"
      },
      {
        "repo": "openai/model_spec",
        "commit": "7f1cf79fcb656c07f77c8d95b6fbc78dc7fac5b6",
        "date": "2026-08-17",
        "verified": false,
        "reason": "unsigned",
        "url": "https://github.com/openai/model_spec/commit/7f1cf79fcb656c07f77c8d95b6fbc78dc7fac5b6"
      }
    ]
  },
  "interpretation": {
    "anthropic": {
      "safety_governance_public_signal": "high",
      "operational_controls_public_signal": "medium",
      "developer_commercial_public_signal": "high",
      "provenance_public_signal": "medium_high_sampled",
      "rationale": "Constitutional AI, Constitution, versioned RSP, alignment/research work, evals, SDKs, plugins, and explicit plugin/action trust warnings are all public. Policies are voluntary and sampled, not exhaustive."
    },
    "openai": {
      "safety_governance_public_signal": "high",
      "operational_controls_public_signal": "medium_high",
      "developer_commercial_public_signal": "very_high",
      "provenance_public_signal": "mixed_sampled",
      "rationale": "Charter, Preparedness Framework, Model Spec, frontier evals, guardrails, privacy/security tooling, SDKs, agents, Cookbook, and open-weight work are all public. Sampled tip signatures are mixed."
    },
    "conclusion": "Public evidence supports mixed incentives and outputs: both organizations invest in safety/ethics/catastrophic-risk governance and in commercialization/developer adoption. Git history cannot establish private motive, profit allocation, deployment effectiveness, or that existential risk has been solved."
  },
  "sources": {
    "anthropic": [
      "https://www.anthropic.com/research/constitutional-ai-harmlessness-from-ai-feedback",
      "https://www.anthropic.com/constitution",
      "https://www.anthropic.com/responsible-scaling-policy",
      "https://www.anthropic.com/research"
    ],
    "openai": [
      "https://openai.com/charter/",
      "https://openai.com/index/updating-our-preparedness-framework/",
      "https://openai.com/safety/",
      "https://github.com/openai/model_spec"
    ]
  }
}