The organization has teamed up with Pangram to introduce innovative features.
Substack has introduced a new tool designed to identify AI-generated content, created in collaboration with Pangram. This feature enables readers to analyze Substack articles to determine the proportion of text generated by AI. The tool is applicable to any text longer than 100 words published starting today. Additionally, Substack is introducing a new section for creators to disclose how AI was utilized in their content. The AI detection feature is currently available on web and iOS platforms, with plans for Android support in the future.
The announcement blog post has a notably sharp tone, making a jab at LinkedIn for its AI-generated content and coining the term “Claudefishing” to describe the artificial attempts at human engagement. However, this critique could be a gamble, as even the most sophisticated AI detection tools cannot guarantee accurate results. A deeper examination into Pangram’s effectiveness as an AI detection tool reveals that it is not foolproof.
In its announcement, Substack recognized the limitations of Pangram’s detection capabilities and suggested that more features related to AI content and user preferences might be in development. The company emphasized that these measures are designed to clarify expectations for readers, encapsulating its position as “people should know what they’re getting.”
