Researcher Questions Pangram Detector Over Diary False Positives
Calls for sharing actual text, writing process, and pangram link instead of personal documents.
Vie McCoy, an OpenAI researcher focused on AI futures, replied to Nathan Worsley about the Pangram AI detector. McCoy asked why Worsley does not share non-personal text along with the writing process and the pangram link. The reply states it is strange that reported false positives involve documents people are not comfortable sharing. The post does not include any specific examples, links, or confirmed detector results.
Combined views
28.7K
3 posts, first seen 2d ago
Reactions
3 postsV⟢ 
vie ⟢@viemccoy@NathanWorsley_ Why don't you write something non-personal and share the actual text and writing process as well as the pangram link? It's really strange that all the false positives are documents that people arent comfortable sharing!
- likes: 89
- replies: 1
- bookmarks: 0
- reposts: 1
BR Brian Roemmele@BrianRoemmeleThe Pentagram “AI Detector” claims your personal diary as—AI. Not just some words—but 1000% of the words. See how this grift works? You no longer have your own words, they are now in control of the Slop Janitor.
- likes: 229
- replies: 27
- bookmarks: 14
- reposts: 35
SH shaur@shaurizardinsane how every single post you see refuting pangram either - can't share the source because it's a personal diary - was using some other site by accident - had ai rewrite a portion of the text by accident idgi is there some sort of defamation campaign from humanizers or smtg
- likes: 275
- replies: 15
- bookmarks: 11
- reposts: 10