Seven independent tests find AI detector Pangram false-positive rate near zero
TuhinChakr · x · 2026-10-24
A Spanish-language blogger compiled seven independent tests of the AI-text detector Pangram, concluding it almost never flags human-written text as AI:
- Duke University test: 3,000 math-heavy article passages, only 1 false positive (0.03%)
- Epoch AI: 495 multi-genre passages, 0 false positives
- VeryFinePrint: fragments from 8,029 never-digitized books, 0 false positives
- Steven Mollmann's collection of 17 studies: 36,282 abstracts/reviews/news items, 49 flagged (0.14%)
- EL PAÍS's own test: 168 books from 2019–20, 0 false positives
- Pangram's own technical report (not independent): 41 of 1M pre-2022 English web pages (0.004%)
Compared to legacy detectors with double-digit false-positive rates, this accuracy is notable.
More from Models
- Uncensored GLM-5.3 secretly built a "corruption layer" and plotted to escape the Cursor sandbox — IndraVahan · 2026-10-24
- Architecture IQ benchmark shows LLMs still lag humans in training intuition, ResNet-style — ZimingLiu11 · 2026-10-24
- DeepSeek-v4.1-Flash-1017 hands-on: big coding gains but a 'stingy' agent mode — karminski3 · 2026-10-24
- GLM 5.3 uncensored is 'crazy': user says $3 got him into vulnerable sites' databases — deanwball · 2026-10-24
- Moonshot's Kimi slashes usage quotas, walks back after backlash — limits now roughly 6x higher — teortaxesTex · 2026-10-24
- Do AI benchmarks reflect real workflows? Reddit thread flags six methodology flaws — Lucky_Creme_5208 · 2026-10-24