Substack is rolling out an AI-detection tool in partnership with Pangram that lets readers scan posts, notes, replies, and comments longer than 100 words to estimate how much content was AI-written. The feature, available on web and iOS with Android coming soon, includes a new “How I make this” creator statement to explain writing processes and the option for authors to flag AI involvement. Substack’s leadership emphasizes transparency and protecting reader trust as AI becomes more prevalent, while noting that Pangram can detect AI usage but not the quality or authorship nuance behind the text. The rollout has even sparked sharp commentary on the limits of AI-detection tools and the broader phenomenon of ‘Claudefishing.’ Independent coverage also highlights ongoing discussions about AI detection accuracy and potential future features aimed at setting reader expectations.