100 classic texts tested: Pangram flags Victor Hugo poem as 100% AI after one line removed
GolinoHudson · x · 2026-10-08
An academic reports that Dartmouth's College of Arts and Sciences now recommends AI detector Pangram, yet his experiments reveal fragility: a pasted copy of Victor Hugo's "Demain, dès l'aube" was flagged 100% AI while the original scored 100% human—differing only in straight apostrophes, collapsed line breaks, and a missing last line. He then ran 100 public-domain classics in 7 languages (Shakespeare, Keats, Dickinson, Hugo, Baudelaire, Goethe, Dante, KJV...), 5 versions each, through Pangram's API to systematically test how punctuation and formatting shifts affect verdicts.
More from Models
- How Long Before Anthropic Ships a Decision Model Like Jev? — tristanbob · 2026-10-08
- Reddit user spots unconfirmed Gemini 4.1 appearing in the app, slow for a Flash model — iamsultanakbar · 2026-10-08
- Jev hits 32% of OpenRouter classification requests in 2 weeks; orgs now juggle 12 models — Tiancaixinxin · 2026-10-08
- Liquid AI open-sources d1 decision models: 3B model answers in one forward pass, 8ms on RTX 4090 — huggingface · 2026-10-08
- Hands-on: OpenAI's new Decisions API edges out Jev in accuracy and image classification — cantrell · 2026-10-08
- New Haiku matches last year's frontier, meaning Claude Code-class capability is heading toward 'too cheap to meter' — johnvmcdonnell · 2026-10-08