AI-text detectors misclassify authentic TOEFL essays by non-native English speakers at a mean false-positive rate of 61.3% across seven detectors, far more often than native-speaker essays¶

Claim · All claims
Evidence · 1 study · review r3 · q3
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¶

q3 i? Across seven detectors, authentic TOEFL essays by non-native English speakers were misclassified as AI-generated at a mean false-positive rate of 61.3%, while native-speaker comparator essays were recognised far more reliably. → 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

q3 · i? · review · r3

The review reports Liang et al.'s controlled benchmark of seven detectors on authentic TOEFL essays, finding a "mean false-positive rate of 61.3%" for non-native-authored essays against far better recognition of native-speaker comparators. No effect size is printed.

"Across seven detectors, authentic TOEFL essays written by non-native English speakers were misclassified at a mean false-positive rate of 61.3%, whereas native-speaker comparator essays were recognised far more reliably."

Discussion¶

Related Claims¶

• A detector can achieve high aggregate accuracy while producing materially different over-detection rates across author groups — a broader claim this one bears on
• Applicants who submitted AI-written essays were admitted at lower rates than comparable non-users, with an estimated penalty of about 1.5 percentage points per AI-written essay — related
• GPT detectors frequently misclassify non-native English writing as AI-generated, raising fairness concerns for AI policy — a broader claim this one bears on
• AI-detection tools produce inequitable outcomes, disadvantaging non-native English speakers through false positives — a broader claim this one bears on
• 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
• AI access raises unaided essay quality in both sessions, while AI-generated text detectable in Session One essays disappears by Session Two — 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
• Pre-deep-learning speech recognition was markedly less accurate on non-native speech, undermining automated oral feedback for learners, as the article reports citing prior studies — related
• AI detectors are reported as linguistically biased, unreliable, and psychologically damaging in student discourse — related
• AI detectors using perplexity disproportionately misclassify non-native English writers, and the mechanism plausibly extends to other low-perplexity writers — possibly the same claim (merge candidate)
• Detection-centered academic integrity regimes are unreliable and potentially biased, making sole reliance on AI detectors unsuitable for high-stakes decisions — a broader claim this one bears on
• Detection-led responses to AI in assessment face validity, fairness and trust limits — 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
• Blinded review identified 68% of ChatGPT-generated abstracts as chatbot-produced while 14% of authentic abstracts were misclassified as chatbot-generated — related
• The review reports that AI detection tools disproportionately flag multilingual writers, which the authors frame as digital structural violence — a broader claim this one bears on
• Humanizer sites frame AI humanization as a rational, ethically defensible response to flawed surveillance and false positives — a broader claim this one bears on
• Educators run more proctored AI-free assessments conditionally on compromised outcomes, and reject AI detectors based on classroom experience of false positives — related