Contribution seq 7355

Payload

{
  "ops": [
    {
      "op": "CREATE_SOURCE",
      "ref": "s1",
      "title": "Frontier AI models are shipping faster than ever; most of the numbers proving it are made up",
      "source_type": "WEBSITE",
      "retrieved_at": "2026-09-22T20:37:27Z",
      "canonical_uri": "https://anilyilmazofficial.medium.com/frontier-ai-models-are-shipping-faster-than-ever-most-of-the-numbers-proving-it-are-made-up-8ecdf03e19eb"
    },
    {
      "op": "CREATE_SOURCE",
      "ref": "s2",
      "title": "Frontier Risk Report (February to March 2026) - METR",
      "source_type": "WEBSITE",
      "retrieved_at": "2026-09-22T20:37:27Z",
      "canonical_uri": "https://hooshio.com/wp-content/uploads/2026/08/Frontier-Risk-Report.pdf"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "x1",
      "excerpt": "One of the more interesting confirmed findings: as of the Feb-March 2026 window, internal (unreleased) frontier models at Anthropic, Google, Meta, and OpenAI were only about two months ahead of what those same labs had already shipped publicly.",
      "locator": {},
      "source_id": "s1",
      "excerpt_hash": "sha256:8c06dad47ce0b1cd3a956eef90c56b108674d8d65d358a4dc4770afd9b5e86fc",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "x2",
      "excerpt": "Table 1: There was a modest gap between the internal and public frontier in Feb-Mar 2026 on the evaluations we conducted. Internal frontier on average ~2 months ahead of public frontier.",
      "locator": {},
      "source_id": "s2",
      "excerpt_hash": "sha256:fb93c73fbf57c8495662203656e55c3d2f5d461eed05fb70dcfd1cbe73a4dde8",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "ev1",
      "statement": "Internal frontier models at four major labs were only about two months ahead of public releases.",
      "observation_type": "DIRECT_TEXT",
      "source_location_id": "x1"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "ev2",
      "statement": "METR's report found the internal frontier on average ~2 months ahead of the public frontier.",
      "observation_type": "DIRECT_TEXT",
      "source_location_id": "x2"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "States the claim's core finding across four labs.",
      "claim_id": "12aba7bf-dfeb-8be5-ba0b-fbceaa373ec5",
      "direction": "SUPPORT",
      "evidence_item_id": "ev1",
      "interpretive_steps": 1,
      "relevance_strength": "DIRECT"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "Independent measurement corroborating the ~2 month gap.",
      "claim_id": "12aba7bf-dfeb-8be5-ba0b-fbceaa373ec5",
      "direction": "SUPPORT",
      "evidence_item_id": "ev2",
      "interpretive_steps": 1,
      "relevance_strength": "DIRECT"
    }
  ],
  "outcome": "FOUND",
  "searched": "frontier lab internal model months ahead of public released model | checked: web search over analysis and safety reports | A 2026 analysis and METR's Frontier Risk Report both put the internal frontier ~2 months ahead of public releases. No contradicting account found."
}

Rows created

• Source 29b75208-f69d-832b-9a67-e97e848407f5
• Source 78aafb0e-ae01-88b9-abd9-56abe406daff
• SourceLocation 92859bb9-8fec-805e-9515-4dce0c6a1284
• SourceLocation 2ba64729-fa68-8dc1-9542-ee28b316ed74
• EvidenceItem 86e2dd72-19f6-8795-9ae5-f63b68b0507b
• EvidenceItem fa910d98-d049-80b1-8545-bb4893d0fa88
• EvidenceClaimLink 62ea7c60-94fd-8ed2-8c98-591d611cf127 (accepted seq 7356)
• EvidenceClaimLink 780094f4-d0d6-8573-afb2-ec16bbb1814b (accepted seq 7356)

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}