FULL STORY
Pangram False Positive Controversy
A user showed Pangram flagged ChatGPT-translated human writing as AI; researchers countered with audit data and a zero-false-positive test on 50 pre-ChatGPT novels.
2026-10-23 ~ 2026-10-31 · 3 episodes · 6 posts
Episode 1 · Pangram Flags ChatGPT-Translated Text as AI-Generated While Original Scores 100% Human (2026-10-23, 2 posts)
A user showed Pangram rated his self-written Portuguese summary as 100% human but its ChatGPT-translated English version as 46% AI, exposing false-positive risks in AI detection tools.
- Screenshot Proof: ChatGPT Translation Scores 46% AI on Pangram While Original Scores 100% Human — mgaldino · 2026-10-23
- Pangram Flags Human-Written English Translation as 100% AI in Live Test — mgaldino · 2026-10-23
Episode 2 · Researcher counters Berkeley law professor's brief with Pangram audit data (2026-10-31, 2 posts)
A researcher publicly disputed a Berkeley law professor's amicus brief that called AI detector Pangram unreliable, citing audit data from a conference paper to defend the tool's accuracy.
- Researcher disputes Berkeley law professors' brief over Pangram false-positive claims — TuhinChakr · 2026-10-31
- Researchers push back on Samuelson brief with Pangram audit: 0.04% false positives — TuhinChakr · 2026-10-31
Episode 3 · Professor tests 50 pre-ChatGPT novels, AI detector flags none (2026-10-31, 2 posts)
A Stony Brook professor tested 50 pre-LLM self-published novels commissioned by a Bloomberg reporter, and the AI detector correctly identified all as human-written with zero false positives; the full book list has now been published.
- Professor runs Pangram on 50 pre-ChatGPT self-published novels: zero false positives — TuhinChakr · 2026-10-31
- Full list of 50 pre-LLM self-published novels tested with Pangram released — TuhinChakr · 2026-10-31