Tool to Detect LLM Clichés

Simon Willison · rss · 2026-07-17

The author built an "LLM cliché highlighter" tool designed to flag common writing tropes found in AI-generated text. The motivation stemmed from fatigue toward formulaic articles packed with repetitive phrases like "no fluff, no filler, no jargon."

This tool was rapidly vibe-coded using Fable 5. It aims to highlight about ten common LLM writing patterns, helping readers quickly identify paragraphs that "look like they were written by AI." The author categorizes it under tools / ai / generative-ai / llms.

Original post →

More from Apps

Apps channel →