Updated
Updated · TechCrunch · Jul 22
Substack Adds Pangram AI Detector for Posts Over 100 Characters as It Pushes Transparency
Updated
Updated · TechCrunch · Jul 22

Substack Adds Pangram AI Detector for Posts Over 100 Characters as It Pushes Transparency

3 articles · Updated · TechCrunch · Jul 22

Summary

  • Substack said users can now scan posts, comments and replies in its app to estimate how much text was written by humans versus AI.
  • The Pangram integration applies to any item above 100 characters and is paired with an optional AI Author’s note so writers can disclose how they used AI.
  • Substack said the tool is meant to encourage transparency rather than punish AI-assisted writing; publishers can also scan drafts before publication and challenge mistaken results.
  • CEO Chris Best framed the feature as a way to preserve trust and keep "AI slop" off the platform, even if exposing AI-written newsletters could hurt Substack’s reputation in the short term.
  • The move extends a broader industry shift toward labeling AI-made content across social media, video, photos and music services.

Insights

With new laws looming, is Substack’s AI scanner a real trust-builder or just a temporary fix?
As AI detectors police platforms, could writers face a new 'cancel culture' for their style?