Substack has launched a new feature that integrates with the AI writing detection software Pangram to give users an estimate of how much content in posts, comments, and replies was written by AI. The tool is now available in the Substack app for any content over 100 characters. In the short term, this transparency could hurt Substack’s business by exposing newsletters that rely heavily on AI, potentially eroding trust in its ecosystem of independent writing. But the company frames it as a long-term bet: keeping the platform free of “AI slop” and encouraging readers to trust that what they read was written by a person, or at least to understand when it is not.

Substack joins a broader industry trend of labeling AI-generated content. Social media platforms label AI-generated photos and videos, and music streaming services have begun penalizing AI-generated music. Substack CEO Chris Best explained the philosophy in a chat with Pangram founder Max Spero: “Software should do everything else, but I think you do want the person to do the hard part” — meaning the original idea worth sharing. The feature is not meant to prohibit or penalize AI-assisted writing. Instead, Substack encourages writers to add an optional “how I make this” statement to properly disclose their use of AI. Publishers can run Pangram on their own drafts before publication, and they can report and remove scans on their own work if they believe the detection is a mistake.

The article notes that the tool provides an estimate, not an absolute verdict, and Substack clarified that it is about transparency, not enforcement. The move could affect how writers and readers perceive authenticity on the platform, but the company is betting that long-term trust outweighs short-term exposure of AI-written content.