Substack adds AI-use scans for posts, notes and comments

Substack is adding a way for readers to check whether writing on the platform appears to be human-written or AI-generated. For anyone paying for newsletters or building an audience there, the feature puts authorship and disclosure closer to the center of the product.

The company said this week that it is integrating Pangram, an AI writing detection tool, into Substack’s app. The feature will let users scan eligible posts, notes, replies and comments to see an estimate of how much of the text was written by a person versus artificial intelligence.

AI detection is not the same as proof. These tools look for patterns in writing and return an estimate, which means the result can be disputed. Substack told TechCrunch that writers will be able to report scans on their own work that they believe are wrong and remove them.

How the feature works

Substack said the scanner will be available inside its app for posts, notes, replies and comments that are longer than 100 characters. The company is also giving publishers access to Pangram before they publish, so writers can check drafts themselves.

Substack is also adding an optional AI Author’s note, according to TechCrunch. That note gives creators a place to disclose whether they used AI and describe how they made the piece.

The company told TechCrunch the feature is not designed to ban AI-assisted writing or punish people who use it. Substack said the goal is to push writers toward more disclosure, including a statement explaining their process.

Why Substack is doing this now

The move arrives as platforms are under pressure to show users when content was made with generative AI, which refers to software that can create text, images, video or audio from prompts. TechCrunch noted that social platforms have already started labeling AI-made photos and videos, while music services have begun labeling some AI-generated tracks and, in some cases, limiting monetization around them.

For Substack, the tension is clear. The platform is built around paid and free relationships between individual writers and readers. If readers believe a newsletter is mostly automated, that could affect trust in the writer and the platform. On the other hand, clearer disclosure could help readers judge what they are paying attention to and what they are paying for.

Substack CEO Chris Best framed the Pangram integration as a useful role for AI. In an online chat with Pangram founder Max Spero, Best said software should handle much of the work around publishing, while the writer still needs to provide “an idea that’s worth reading, that’s worth caring about, that’s worth sharing.”

Best added in that conversation that “you do want the person to do the hard part.”

The feature gives Substack a more visible stance in the debate over AI-written media: AI can be part of the workflow, but readers should get more information about where the human work ends and automation begins.

This story draws on original reporting from TechCrunch.