Substack has introduced a new AI detection feature that allows readers to check whether posts, Notes, replies and comments were written by humans or generated with the help of AI.

How the Tool Works

Powered by AI detection company Pangram, the feature allows users to scan eligible content by selecting the three-dot menu and choosing "Scan for AI text." The tool then estimates what percentage of the text is AI-generated, AI-assisted or human-written.

The feature is currently available on Substack's web platform and iOS app, with Android support expected to follow.

However, the tool only works on text passages longer than 100 words and content published after July 21, 2026. Substack says the detection only happens when a reader actively requests a scan.

Creators Can Explain or Disable Detection

Substack is also giving writers more control over how their use of AI is presented. Creators can add a statement explaining how they produced their content, which readers can see when they run an AI scan.

Writers can also scan their own drafts and submit a report if they believe Pangram has incorrectly classified their work. More notably, creators can disable AI detection on their content entirely, removing the ability for readers to run the scan.

Substack CEO and co-founder Chris Best said the feature is intended to address situations where readers may assume they are engaging with human-written work when AI has played a significant role in producing it.

The move reflects a growing debate across digital publishing about AI-generated content, transparency and audience trust, as platforms look for ways to help readers understand how the content they consume is created.