Substack has launched a new AI detection tool in partnership with Pangram to assess how much content was written by artificial intelligence. The tool, available for text longer than 100 words published from today, can be used on web and iOS, with Android support forthcoming. This feature aims to help readers understand the origin of the content they consume.

In addition to detection, Substack is introducing a new statement space for creators to explicitly disclose their use of AI in content creation. The company's announcement includes playful jabs at other platforms, coining the term "Claudefishing" for attempts at feigned human connection with AI-generated text.

Substack acknowledges the limitations of current AI detection technology, noting that even the best tools aren't foolproof. The Atlantic previously explored the accuracy issues with AI detection tools, including Pangram. The company emphasizes that these new measures are primarily about managing reader expectations, stating, "people should know what they're getting."