Substack partners with Pangram to introduce AI content transparency tool

Substack has launched a new artificial intelligence detection feature developed in partnership with Pangram, marking a significant move towards content transparency on the newsletter platform. The tool enables readers to assess the proportion of AI-generated material in posts that exceed 100 words. This initiative is designed to provide greater clarity regarding the origins of content, allowing subscribers to better understand what they are reading.

The detection capabilities are currently available on web and iOS platforms, with support for Android devices scheduled for release in the near future. Alongside the detection tool, Substack has introduced a dedicated space for creators to explicitly disclose their use of artificial intelligence in their work. This dual approach aims to set clear expectations for readers while providing a structured mechanism for creators to maintain openness about their production processes.

In its announcement, Substack framed the initiative as a measure to ensure that people know what they are getting, emphasising the importance of transparency. However, the company acknowledged the inherent limitations of the technology, recognising that current detection tools cannot guarantee perfect accuracy. This cautious stance aligns with broader industry observations, including previous investigations by The Atlantic, which have highlighted that AI detection tools, including Pangram, are far from perfect and cannot definitively confirm the origin of all content.

The launch announcement also featured a pointed critique of competitors, specifically targeting LinkedIn. Substack described the prevalence of AI-generated content on that platform as "AI slop" and coined the term "Claudefishing" to describe attempts to create feigned human connections using artificial intelligence. While this editorial stance highlights Substack’s position on content quality, it also underscores the subjective nature of such labels and the ongoing debate surrounding AI in professional networks.

Looking ahead, Substack hinted at additional features it is considering regarding AI content and user preferences. By combining technical detection with mandatory disclosure spaces, the platform is attempting to navigate the complex landscape of generative AI. The move reflects a growing demand for accountability in digital publishing, even as the underlying technology continues to evolve and face scrutiny regarding its reliability.