Against Claudefishing – AI detection feature on Substack (post.substack.com)

Substack has announced a new AI detection feature in collaboration with Pangram, enabling users to scan text to determine whether it was generated by human writers or AI. This initiative addresses the growing concern of "Claudefishing," where content produced by AI may mislead readers expecting authentic human insight. With estimates suggesting up to 40% of text on some social media platforms is AI-generated, Substack aims to foster a transparent environment that maintains trust between content creators and readers. The feature, available on both web and iOS with plans for Android, assesses text longer than 100 words, offering insights only to those seeking them. While it accurately detects the involvement of AI, this tool doesn't assess the quality or effort behind the writing. In addition to the detection capabilities, Substack introduces resources for creators to better communicate their writing processes, helping establish clearer expectations for readers. This move is significant for the AI/ML community, as it emphasizes the importance of transparency in content creation and underscores a collaborative approach to navigating the evolving landscape of human and AI-generated content.