In a documented exploratory audit, one human-authored manuscript received materially different classifications from five commercial detectors, spanning 0% human to Human Generated¶

Claim · All claims
Evidence · 1 study · review r3 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? The same known-provenance manuscript, submitted to five commercial detectors on the same day, received classifications ranging from Winston AI's 0% human to GPTZero's categorical label Human Generated, with three AI-leaning and two human-leaning outcomes. → Victor Angelier 2026

Evidence¶

Victor Angelier 2026¶

Victor Angelier. (2026). The pitfalls of AI detection in academic writing: bias, false positives, and the need for inclusive assessment. AI and Ethics. https://doi.org/10.1007/s43681-026-01376-w

q2 · i? · review · r3

The author's documented exploratory audit of one manuscript (4,613–4,727 words) on 30 August 2026, with timestamped, hash-verifiable artefacts in Supplementary Data S1, found "substantial cross-tool inconsistency". The audit was not blinded and used single runs; it supports existence claims, not prevalence estimates.

"Winston AI reported 0% human. Copyleaks and Originality.ai returned approximately 80–81% AI. GPTZero returned 42% AI while assigning the categorical label "Human Generated". Tool E reported 34% AI / 66% human."

Discussion¶

Related Claims¶

• Universities disabling AI detection cite false-positive risk at scale, and detector influence persists in individual grading, journal screening, and institutions without policy revision — related
• Independent evaluation found commercial AI-text detectors neither sufficiently accurate nor sufficiently reliable for high-stakes use — a broader claim this one bears on
• In a paired exploratory comparison, extensive machine rewriting by a commercial humanizer materially degraded the manuscript while changing the detection score only slightly — related
• Admissions officers can often discriminate AI-written from human-written essays, achieving an AUC of 0.70, well above chance but below commercial detectors — related
• AI language polishing alone can flip human-authored scholarly texts from human-classified to AI-classified — related
• GPT detectors frequently misclassify non-native English writing as AI-generated, raising fairness concerns for AI policy — related
• AI-detection tools produce inequitable outcomes, disadvantaging non-native English speakers through false positives — related
• AI access raises unaided essay quality in both sessions, while AI-generated text detectable in Session One essays disappears by Session Two — related
• Of 100 human-audited claims, 39 are fully verifiable, 55 partially verifiable, and 6 not verifiable, with main responses showing stronger provenance than follow-ups — related
• AI text detectors can return high AI-detection scores for human-written academic text, raising fairness concerns about authenticity evaluation — related
• AI-content detection tools achieve only marginally higher success than human reviewers, around 80% versus 78.4% — related
• AI-generated text can pass as human-authored in up to 80% of cases, making reliable differentiation a significant challenge — related
• Humanizer sites frame AI humanization as a rational, ethically defensible response to flawed surveillance and false positives — related
• All four detection tools classified 100% of fully human-written papers as true negatives, with false positives nearly absent — reports the opposite
• Comparative studies report that AI detection tools do not reliably distinguish human-written from AI-generated text — a broader claim this one bears on
• AI-generated-text detectors are reported as ineffective against current sophisticated language models, and students can alter text to evade detection — related
• OpenAI's GPT-2 Output Detector distinguished real student essays from ChatGPT-generated essays on the same subjects in a December 2022 trial — related
• Blinded review identified 68% of ChatGPT-generated abstracts as chatbot-produced while 14% of authentic abstracts were misclassified as chatbot-generated — related
• Educators run more proctored AI-free assessments conditionally on compromised outcomes, and reject AI detectors based on classroom experience of false positives — related