{
  "study": "The AI detector false-accusation ledger: every documented case, detector and outcome",
  "url": "https://provenbrief.com/story/the-ai-detector-false-accusation-ledger-every-documented-case-detector-and-outco",
  "publisher": "ProvenBrief",
  "contentUpdatedAt": "2026-11-10T01:59:36.552Z",
  "terms": "Free to use with attribution to ProvenBrief and a link to the source page.",
  "tables": [
    {
      "key": "fp-evidence",
      "title": "Claimed vs measured false-positive rates, AI writing detectors",
      "measure": "share of tested human-written writing that the detector or study labeled as AI-generated",
      "unit": "% of essays falsely flagged",
      "rows": [
        {
          "key": "turnitin-claimed",
          "label": "Turnitin's claimed false-positive rate, 2023",
          "value": 1,
          "note": "Chechitelli claim via Vanderbilt statement; The Register prints 'less than one percent'",
          "source": "https://www.vanderbilt.edu/brightspace/2023/08/16/guidance-on-ai-detection-and-why-were-disabling-turnitins-ai-detector/"
        },
        {
          "key": "nonnative-average",
          "label": "Average across 7 detectors, non-native English essays (Patterns, 2023)",
          "value": 61.2,
          "note": "arXiv v3 prints 61.22%; prose and table round to 61.2; the seven detectors excluded Turnitin",
          "source": "https://arxiv.org/html/2304.02819v3"
        },
        {
          "key": "wapo-test",
          "label": "Washington Post test of Turnitin, 2023",
          "value": 50,
          "note": "smaller sample, via Wikipedia citing the Washington Post; tested Turnitin itself",
          "source": "https://en.wikipedia.org/wiki/Artificial_intelligence_content_detection"
        },
        {
          "key": "csm-black",
          "label": "Common Sense Media 2024, writing by Black students",
          "value": 20,
          "note": "via Wikipedia citing the Common Sense Media report",
          "source": "https://en.wikipedia.org/wiki/Artificial_intelligence_content_detection"
        },
        {
          "key": "csm-latino",
          "label": "Common Sense Media 2024, writing by Latino students",
          "value": 10,
          "note": "via Wikipedia citing the Common Sense Media report",
          "source": "https://en.wikipedia.org/wiki/Artificial_intelligence_content_detection"
        },
        {
          "key": "csm-white",
          "label": "Common Sense Media 2024, writing by White students",
          "value": 7,
          "note": "via Wikipedia citing the Common Sense Media report",
          "source": "https://en.wikipedia.org/wiki/Artificial_intelligence_content_detection"
        }
      ]
    },
    {
      "key": "detector-tally",
      "title": "Which tools the documented 2023 false accusations named",
      "measure": "case clusters in this ledger where the reporting names the detection tool",
      "unit": "documented case clusters",
      "rows": [
        {
          "key": "turnitin",
          "label": "Turnitin",
          "value": 2,
          "note": "assembled across ledger rows: Johns Hopkins (Markup) and UC Davis Stivers (Futurism)",
          "source": "https://themarkup.org/machine-learning/2023/08/14/ai-detection-tools-falsely-accuse-international-students-of-cheating"
        },
        {
          "key": "gptzero",
          "label": "GPTZero",
          "value": 1,
          "note": "UC Davis Quarterman case; Rolling Stone names the GPTZero flag",
          "source": "https://www.rollingstone.com/culture/culture-features/student-accused-ai-cheating-turnitin-1234747351/"
        },
        {
          "key": "chatgpt-as-detector",
          "label": "ChatGPT itself, used as a detector",
          "value": 1,
          "note": "Texas A&M University-Commerce class episode",
          "source": "https://en.wikipedia.org/wiki/Artificial_intelligence_content_detection"
        }
      ]
    },
    {
      "key": "retreat-actions",
      "title": "Public retreats from AI writing detection, 2023",
      "measure": "withdrawals, shutdowns and institutional opt-outs documented in reporting or statements",
      "unit": "documented public actions",
      "rows": [
        {
          "key": "vendor-shutdowns",
          "label": "Detector vendors that shut their own tools down",
          "value": 2,
          "note": "OpenAI's classifier (end of July 2023) and the Quill.org/CommonLit AI Writing Check",
          "source": "https://themarkup.org/machine-learning/2023/08/14/ai-detection-tools-falsely-accuse-international-students-of-cheating"
        },
        {
          "key": "university-withdrawals",
          "label": "Universities that publicly disabled or withdrew support",
          "value": 4,
          "note": "Cambridge, Pittsburgh, Vanderbilt, UT Austin (Sept 2023, five months after Cambridge)",
          "source": "https://en.wikipedia.org/wiki/Artificial_intelligence_content_detection"
        },
        {
          "key": "turnitin-optout",
          "label": "Vendor that shipped an institutional opt-out switch",
          "value": 1,
          "note": "Turnitin, June 2023; two percent of customers had flipped it by end of July, per the company",
          "source": "https://themarkup.org/machine-learning/2023/08/14/ai-detection-tools-falsely-accuse-international-students-of-cheating"
        }
      ]
    }
  ]
}