AI Text Detector Pangram Shows No False Positives But Fails Against Modern LLMs
FlorianGallwitz · x · 2026-08-05
The author shares interim results from a study on the AI text detector Pangram. It shows virtually no false positives, but a good modern LLM can easily generate texts that Pangram classifies as human-written.
More from Models
- User Complaints: Single Task Burns Through a Week's Worth of Grok Quota — huangyun_122 · 2026-08-05
- lightonai Open-Sources Training and Eval Scripts for mDenseOn and mLateOn Retrievers — IgorCarron · 2026-08-05
- Anthropic Cut 80% of Claude Code's System Prompt with No Measurable Eval Loss — bibryam · 2026-08-05
- Local Full-Precision 27B Beats Heavily Quantized Flagship in Terminal-Bench — Saber-tooth-tiger · 2026-08-05
- Kimi Hailed as the New Claude, Moonshot as the New Anthropic by Creatives — EXM7777 · 2026-08-05
- Claude Opus 5 Overuses 'silently' and 'load-bearing', Data Shows — JeremyNguyenPhD · 2026-08-05