Writers add typos and clumsy phrasing to avoid AI-detection suspicion
Overview
Writers now deliberately insert typos, broken syntax and profanity to make their text look human, QbitAI reports, because polished prose is increasingly flagged as machine-written.
QbitAI cites a Pew analysis of 490,000 web pages published before ChatGPT, in which over a third showed clear AI-editing traces. It also reports that Mia Ballard's novel was flagged at 78.3% AI writing, after which Hachette canceled its US edition. The report does not include a detector's accuracy rate or Hachette's stated reason for the cancellation.
Written by AI from the articles below · updated Nov 7, 6:15 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- QbitAINewsWriters deliberately add typos and clumsy phrasing to avoid AI-detection suspicion
AIQbitAI argues that writers now deliberately include typos, broken syntax, and profanity to make their text look human, because AI-detector suspicion has made polished prose seem machine-written. The article cites Pew's finding that over a third of 490,000 pre-ChatGPT-era web pages it analyzed showed clear AI editing traces, and reports that Mia Ballard's novel was flagged at 78.3% AI writing, after which Hachette canceled its US edition.
Heat Trend
Not enough continuous observations to show a trend yet.