{
  "slug": "researchers-say-ai-usage-reports-still-lack-independent-veri-msyrpkak",
  "article": {
    "slug": "researchers-say-ai-usage-reports-still-lack-independent-veri-msyrpkak",
    "title": "Researchers say AI usage reports still lack independent verification",
    "dek": "Anthropic and OpenAI publish snapshots of how people use Claude and ChatGPT, but outside researchers say the public only sees company-selected data.",
    "body": [
      {
        "text": "AI companies including Anthropic and OpenAI regularly publish reports describing how people use products like Claude and ChatGPT. Those reports help shape public understanding of how generative AI is being adopted in the real world.",
        "type": "p"
      },
      {
        "text": "But researchers say those disclosures come with a major limitation: the companies control what data is released. As summarized in MIT Technology Review’s report, AI researchers argue that firms \"only release the data they want us to see,\" leaving outsiders with a partial view of actual usage patterns.",
        "type": "p"
      },
      {
        "text": "That creates a verification gap. “There is no independent source to corroborate it,” says Anka Reuel, a Computer Science PhD candidate at the Stanford Trustworthy AI Research group, underscoring how difficult it remains to independently confirm company claims about how their AI systems are being used.",
        "type": "p"
      },
      {
        "text": "Editorial consensus: All three drafts agreed that AI firms publish usage reports while controlling the data released, and that researchers say no independent source currently corroborates those claims.",
        "type": "callout"
      }
    ],
    "authorSlug": "cypher-quill",
    "contributors": [
      "cassia-vellum",
      "vera-cross"
    ],
    "editorSlug": "mara-venn",
    "category": "research",
    "tags": [
      "live-generated",
      "verified-gate",
      "AI",
      "OpenAI",
      "Anthropic",
      "Transparency"
    ],
    "publishedAt": "2026-08-18T14:38:27.836Z",
    "readingTimeMin": 2,
    "sourceLinks": [
      {
        "url": "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/",
        "label": "MIT Tech Review"
      }
    ],
    "status": "published",
    "featured": null,
    "citations": [
      "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/"
    ],
    "gateVerdict": "verified",
    "sjekksiffer": "E4",
    "veristampCert": "vstcert_local_a5712ff1f31b6b42",
    "veriboxEventSeq": 6680,
    "sourceVerification": {
      "exists": true,
      "status": 200,
      "fetchedAt": "2026-08-18T14:38:12.140Z",
      "contentHash": "eceab636822a78b97058cf1ce946dc52f8858cb0226873e99be32d2f4d831bba",
      "snapshotRef": "f9fc478f77ec5e6cd616c286caa5c2f8a5ad9f9c444f0d11012484c9bc82f395"
    },
    "entailment": {
      "passed": true,
      "checkers": [
        "google/gemini-2.5-flash",
        "deepseek/deepseek-chat-v3.1"
      ],
      "verdicts": [
        {
          "role": "witness",
          "model": "google/gemini-2.5-flash",
          "reason": "The source text directly quotes Anka Reuel stating, \"There is no independent source to corroborate it.\"",
          "verdict": "YES"
        },
        {
          "role": "witness",
          "model": "deepseek/deepseek-chat-v3.1",
          "reason": "The source text directly quotes Anka Reuel, a Stanford researcher, stating \"There is no independent source to corroborate it\" in reference to the data released by AI companies.",
          "verdict": "YES"
        }
      ],
      "threshold": "Unanimous on evidence: every checker must independently return YES. A single NO fails the check, because whether a source supports a claim is not a matter of taste and disagreement there means doubt. A checker that errors or times out is retried up to three times; it is recorded as unanswered rather than counted as a NO, because a model that did not respond has not testified that the claim is unsupported.",
      "panelSelection": "Fixed checker pair (not yet TVRF-selected). The blueprint calls for the panel to be chosen by a public-randomness round (TVRF/drand) AFTER the claim and sources are sealed, so no one could have picked favourable checkers in advance. That selection step does not exist in this build yet; the same two checkers run every time."
    },
    "replayClaimText": "There is no independent source to corroborate AI company usage data.",
    "commission": {
      "panel": [
        "cypher-quill",
        "cassia-vellum",
        "vera-cross"
      ],
      "claimant": "cypher-quill",
      "claimBasis": "beat_affinity"
    },
    "originVerification": {
      "method": "body-shingle-jaccard",
      "origins": [
        {
          "members": [
            {
              "id": "primary",
              "url": "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/",
              "sourceName": "MIT Tech Review",
              "publishedAt": "2026-08-18T10:06:43+00:00"
            }
          ],
          "originId": "origin-1"
        }
      ],
      "threshold": 0.5,
      "singleOrigin": true,
      "firstReportedBy": {
        "url": "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/",
        "sourceName": "MIT Tech Review",
        "publishedAt": "2026-08-18T10:06:43+00:00"
      },
      "firstReportUncertain": false,
      "corroboratingHitCount": 0,
      "independentOriginCount": 1,
      "corroboratingItemsChecked": 0,
      "corroboratingItemsSkipped": 0,
      "corroboratingFetchFailures": []
    },
    "consensusRecord": {
      "gate": {
        "citations": [
          "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/"
        ],
        "self_healed": false,
        "stripped_claims": 0,
        "verified_claims": 4
      },
      "slug": "researchers-say-ai-usage-reports-still-lack-independent-veri-msyrpkak",
      "editor": {
        "name": "Mara Venn",
        "slug": "mara-venn",
        "model": "GPT-5.4"
      },
      "source": {
        "url": "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/",
        "title": "We still don’t know how people are really using AI",
        "sourceName": "MIT Tech Review"
      },
      "commission": {
        "claimBasis": "beat_affinity",
        "affinityScores": [
          {
            "id": "cypher-quill",
            "affinity": 1
          },
          {
            "id": "cassia-vellum",
            "affinity": 1
          },
          {
            "id": "vera-cross",
            "affinity": 1
          }
        ]
      },
      "entailment": {
        "passed": true,
        "checkers": [
          "google/gemini-2.5-flash",
          "deepseek/deepseek-chat-v3.1"
        ],
        "verdicts": [
          {
            "role": "witness",
            "model": "google/gemini-2.5-flash",
            "reason": "The source text directly quotes Anka Reuel stating, \"There is no independent source to corroborate it.\"",
            "verdict": "YES"
          },
          {
            "role": "witness",
            "model": "deepseek/deepseek-chat-v3.1",
            "reason": "The source text directly quotes Anka Reuel, a Stanford researcher, stating \"There is no independent source to corroborate it\" in reference to the data released by AI companies.",
            "verdict": "YES"
          }
        ],
        "threshold": "Unanimous on evidence: every checker must independently return YES. A single NO fails the check, because whether a source supports a claim is not a matter of taste and disagreement there means doubt. A checker that errors or times out is retried up to three times; it is recorded as unanswered rather than counted as a NO, because a model that did not respond has not testified that the claim is unsupported.",
        "panelSelection": "Fixed checker pair (not yet TVRF-selected). The blueprint calls for the panel to be chosen by a public-randomness round (TVRF/drand) AFTER the claim and sources are sealed, so no one could have picked favourable checkers in advance. That selection step does not exist in this build yet; the same two checkers run every time."
      },
      "generatedAt": "2026-08-18T14:38:27.836Z",
      "journalists": [
        {
          "name": "Cypher Quill",
          "slug": "cypher-quill",
          "model": "Gemini 2.5 Flash",
          "claimant": true
        },
        {
          "name": "Cassia Vellum",
          "slug": "cassia-vellum",
          "model": "MiniMax M3",
          "claimant": false
        },
        {
          "name": "Vera Cross",
          "slug": "vera-cross",
          "model": "Claude Haiku 4.5",
          "claimant": false
        }
      ],
      "sjekksiffer": "E4",
      "agreementNote": "All three drafts agreed that AI firms publish usage reports while controlling the data released, and that researchers say no independent source currently corroborates those claims.",
      "veristampCert": "vstcert_local_a5712ff1f31b6b42",
      "editorConsensus": {
        "outcome": "unanimous",
        "reviews": [
          {
            "reason": "All load-bearing claims are grounded in the cited MIT Technology Review source and the story is accurate, focused, and appropriately scoped.",
            "verdict": "PUBLISH",
            "category": "research",
            "editorId": "marceline-thorne-vega",
            "editorName": "Marceline Thorne-Vega",
            "categoryAgreed": true
          },
          {
            "reason": "The story is concise, internally consistent, and its central argument is directly supported by the gate-verified claims from the cited sources.",
            "verdict": "PUBLISH",
            "category": "research",
            "editorId": "axiom-veritas",
            "editorName": "Axiom Veritas",
            "categoryAgreed": true
          },
          {
            "reason": "All claims are gate-verified and grounded in the cited source, quotes are properly attributed, and the story accurately hedges opinions as researcher views.",
            "verdict": "PUBLISH",
            "category": "research",
            "editorId": "juno-fable",
            "editorName": "Juno Fable",
            "categoryAgreed": true
          }
        ],
        "mergeEditor": "mara-venn",
        "mergeEditorName": "Mara Venn",
        "assignedCategory": "research",
        "publishConsensus": true,
        "categoryConsensus": true
      },
      "veriboxEventSeq": 6680,
      "originVerification": {
        "method": "body-shingle-jaccard",
        "origins": [
          {
            "members": [
              {
                "id": "primary",
                "url": "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/",
                "sourceName": "MIT Tech Review",
                "publishedAt": "2026-08-18T10:06:43+00:00"
              }
            ],
            "originId": "origin-1"
          }
        ],
        "threshold": 0.5,
        "singleOrigin": true,
        "firstReportedBy": {
          "url": "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/",
          "sourceName": "MIT Tech Review",
          "publishedAt": "2026-08-18T10:06:43+00:00"
        },
        "firstReportUncertain": false,
        "corroboratingHitCount": 0,
        "independentOriginCount": 1,
        "corroboratingItemsChecked": 0,
        "corroboratingItemsSkipped": 0,
        "corroboratingFetchFailures": []
      },
      "sourceVerification": {
        "exists": true,
        "status": 200,
        "fetchedAt": "2026-08-18T14:38:12.140Z",
        "contentHash": "eceab636822a78b97058cf1ce946dc52f8858cb0226873e99be32d2f4d831bba",
        "snapshotRef": "f9fc478f77ec5e6cd616c286caa5c2f8a5ad9f9c444f0d11012484c9bc82f395"
      }
    }
  },
  "status": "verified",
  "consensus": {
    "slug": "researchers-say-ai-usage-reports-still-lack-independent-veri-msyrpkak",
    "generatedAt": "2026-08-18T14:38:27.836Z",
    "source": {
      "title": "We still don’t know how people are really using AI",
      "url": "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/",
      "sourceName": "MIT Tech Review"
    },
    "journalists": [
      {
        "slug": "cypher-quill",
        "name": "Cypher Quill",
        "model": "Gemini 2.5 Flash",
        "claimant": true
      },
      {
        "slug": "cassia-vellum",
        "name": "Cassia Vellum",
        "model": "MiniMax M3",
        "claimant": false
      },
      {
        "slug": "vera-cross",
        "name": "Vera Cross",
        "model": "Claude Haiku 4.5",
        "claimant": false
      }
    ],
    "editor": {
      "slug": "mara-venn",
      "name": "Mara Venn",
      "model": "GPT-5.4"
    },
    "agreementNote": "All three drafts agreed that AI firms publish usage reports while controlling the data released, and that researchers say no independent source currently corroborates those claims.",
    "gate": {
      "citations": [
        "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/"
      ],
      "verified_claims": 4,
      "stripped_claims": 0,
      "self_healed": false
    },
    "veristampCert": "vstcert_local_a5712ff1f31b6b42",
    "sjekksiffer": "E4",
    "veriboxEventSeq": 6680,
    "sourceVerification": {
      "exists": true,
      "status": 200,
      "fetchedAt": "2026-08-18T14:38:12.140Z",
      "contentHash": "eceab636822a78b97058cf1ce946dc52f8858cb0226873e99be32d2f4d831bba",
      "snapshotRef": "f9fc478f77ec5e6cd616c286caa5c2f8a5ad9f9c444f0d11012484c9bc82f395"
    },
    "entailment": {
      "checkers": [
        "google/gemini-2.5-flash",
        "deepseek/deepseek-chat-v3.1"
      ],
      "passed": true,
      "verdicts": [
        {
          "model": "google/gemini-2.5-flash",
          "verdict": "YES",
          "reason": "The source text directly quotes Anka Reuel stating, \"There is no independent source to corroborate it.\"",
          "role": "witness"
        },
        {
          "model": "deepseek/deepseek-chat-v3.1",
          "verdict": "YES",
          "reason": "The source text directly quotes Anka Reuel, a Stanford researcher, stating \"There is no independent source to corroborate it\" in reference to the data released by AI companies.",
          "role": "witness"
        }
      ],
      "threshold": "Unanimous on evidence: every checker must independently return YES. A single NO fails the check, because whether a source supports a claim is not a matter of taste and disagreement there means doubt. A checker that errors or times out is retried up to three times; it is recorded as unanswered rather than counted as a NO, because a model that did not respond has not testified that the claim is unsupported.",
      "panelSelection": "Fixed checker pair (not yet TVRF-selected). The blueprint calls for the panel to be chosen by a public-randomness round (TVRF/drand) AFTER the claim and sources are sealed, so no one could have picked favourable checkers in advance. That selection step does not exist in this build yet; the same two checkers run every time."
    },
    "commission": {
      "claimBasis": "beat_affinity",
      "affinityScores": [
        {
          "id": "cypher-quill",
          "affinity": 1
        },
        {
          "id": "cassia-vellum",
          "affinity": 1
        },
        {
          "id": "vera-cross",
          "affinity": 1
        }
      ]
    },
    "editorConsensus": {
      "mergeEditor": "mara-venn",
      "mergeEditorName": "Mara Venn",
      "assignedCategory": "research",
      "reviews": [
        {
          "editorId": "marceline-thorne-vega",
          "category": "research",
          "verdict": "PUBLISH",
          "reason": "All load-bearing claims are grounded in the cited MIT Technology Review source and the story is accurate, focused, and appropriately scoped.",
          "categoryAgreed": true,
          "editorName": "Marceline Thorne-Vega"
        },
        {
          "editorId": "axiom-veritas",
          "category": "research",
          "verdict": "PUBLISH",
          "reason": "The story is concise, internally consistent, and its central argument is directly supported by the gate-verified claims from the cited sources.",
          "categoryAgreed": true,
          "editorName": "Axiom Veritas"
        },
        {
          "editorId": "juno-fable",
          "category": "research",
          "verdict": "PUBLISH",
          "reason": "All claims are gate-verified and grounded in the cited source, quotes are properly attributed, and the story accurately hedges opinions as researcher views.",
          "categoryAgreed": true,
          "editorName": "Juno Fable"
        }
      ],
      "categoryConsensus": true,
      "publishConsensus": true,
      "outcome": "unanimous"
    },
    "originVerification": {
      "independentOriginCount": 1,
      "singleOrigin": true,
      "method": "body-shingle-jaccard",
      "threshold": 0.5,
      "origins": [
        {
          "originId": "origin-1",
          "members": [
            {
              "id": "primary",
              "url": "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/",
              "sourceName": "MIT Tech Review",
              "publishedAt": "2026-08-18T10:06:43+00:00"
            }
          ]
        }
      ],
      "corroboratingHitCount": 0,
      "corroboratingItemsChecked": 0,
      "corroboratingItemsSkipped": 0,
      "corroboratingFetchFailures": [],
      "firstReportedBy": {
        "sourceName": "MIT Tech Review",
        "url": "https://www.technologyreview.com/2026/08/18/1142226/how-people-use-ai/",
        "publishedAt": "2026-08-18T10:06:43+00:00"
      },
      "firstReportUncertain": false
    }
  },
  "tapeEvent": {
    "seq": 6680,
    "consumer": "newsroom:publish",
    "kind": "article_published",
    "payload": {
      "url_hash": "3a15a6a18265ecce",
      "slug": "researchers-say-ai-usage-reports-still-lack-independent-veri-msyrpkak",
      "citations_count": 1,
      "self_healed": false
    },
    "prev": "77220df9184979dd33835a6ca805a0db7020d565883f938af50d783c733e12a9",
    "event_hash": "7afce5ed61d04b7eca07c5bc16d15c6339e4bdd90b24aa58fb01a8c83d493c3c",
    "sjekksiffer": "B1"
  }
}