Substack has launched a feature that identifies which newsletters are being written using artificial intelligence.
The writing platform announced an integration with Pangram, an AI detection service, this week. Users can now scan posts, comments, and replies within the Substack app to see an estimate of how much content was generated by a human versus a machine.
For now, the move could hurt Substack’s business. It exposes the platform to newsletters that rely heavily on automation. This might erode trust in the ecosystem of independent blogs or damage the reputation of the site as a host for high-quality writing.
Longer term, however, the tool could help keep the platform free of low-quality AI content. It encourages users to trust what they read was written by a person, or at least understand when it was not.
Substack joins other platforms that are labelling AI content. Photos and videos created with artificial intelligence are marked on social media sites. Music streaming services have recently begun labelling and, in some cases, penalising AI-generated tracks.
“This is good use of AI,” Substack CEO Chris Best said.
“When I used to pitch Substack to writers, one way I would do it is … we’ll do everything for you except the hard part,” Best explained in an online chat with Max Spero, the founder of Pangram. “You have to have something — an idea that’s worth reading, that’s worth caring about, that’s worth sharing. That one thing is very hard and very valuable … [S]oftware should do everything else, but I think you do want the person to do the hard part.”
The feature is available in the Substack app for any post, note, reply, or comment above 100 characters. Writers can also include an optional AI author’s note to disclose their use of the technology, the company told TechCrunch.
The tool is not meant to prohibit or penalise AI-assisted writing. Instead, it encourages writers to add a statement explaining their process.
Publishers can run Pangram on their own drafts before publication. They can also report and remove scans on their own work if they believe a detection is a mistake.
What it means
Readers now have a way to verify the origin of the text they consume. Writers gain a mechanism to be transparent about their workflow without facing automatic penalties for using automation.
