Substack integrates AI detection to boost transparency in newsletter ecosystem
The platform’s move signals a strategic pivot toward disclosure, aiming to preserve trust among readers while permitting writers to use generative tools for technical tasks.
Substack has introduced a new feature integrated with AI detection software Pangram, enabling users to estimate the proportion of content in newsletters, posts, comments, and replies that is generated by artificial intelligence. The tool is designed to increase transparency regarding AI-assisted writing, allowing readers to distinguish between human-created text and material produced or assisted by software. This development marks a significant step in the platform’s approach to managing the growing role of generative AI in independent publishing.
The feature is available within the Substack app for any post, note, reply, or comment exceeding 100 characters. Substack clarified that the integration is not intended to prohibit or penalise AI-assisted writing. Instead, the company aims to keep the platform free of low-quality content, often referred to as “AI slop,” and to encourage trust within its ecosystem of independent news and blogs. By providing visibility into content origins, the platform seeks to maintain its reputation as a host of high-quality material.
Substack CEO Chris Best described the initiative as a positive application of AI technology. In an online discussion with Pangram founder Max Spero, Best emphasised that while software can handle technical tasks, the core creative idea must remain human. He noted that the value of a newsletter lies in having an idea worth reading and sharing, a task he described as difficult but essential. The company encourages writers to include a “how I make this” statement to explain their creative process, further contextualising their use of technology.
This move aligns with a broader industry shift toward labelling AI-generated content. Social media sites already label photos and videos generated by AI, and music streaming services have begun labeling or penalising AI-generated music. Substack’s decision to adopt similar transparency measures reflects a growing consensus that disclosure is necessary to maintain reader confidence in digital content. The platform joins other services in recognising that while AI can assist in creation, clear attribution is vital for audience trust.
Writers on the platform can optionally include an AI Author’s note to disclose their use of AI tools. Additionally, publishers have the ability to run Pangram scans on their own drafts before publication. If a scan is believed to be inaccurate, publishers can report and remove the scan from their work. This mechanism provides a safeguard against potential errors in detection, acknowledging the ongoing debate regarding the accuracy of AI identification tools.
