Substack Fights AI Deception: New Tool Flags AI-Written Content

Substack is introducing a new tool designed to help readers identify content potentially written by artificial intelligence. Rolling out on the web and iOS, with Android to follow, this feature allows users to scan posts longer than 100 words for AI-generated text. The integration with AI detection company Pangram aims to foster greater transparency on the platform as AI usage becomes more common.

CEO Chris Best emphasized that the tool addresses the 'mismatch between a reader's expectation and reality,' a phenomenon he terms 'Claudefishing.' The platform will also introduce a "How I make this" statement for creators to detail their writing process. While Pangram can only detect AI's involvement, not the human effort or AI's role as a source, Substack's objective is to empower readers to choose what they invest their time in reading, thereby protecting authors' livelihoods.