# Substack adds AI detection to fight ‘Claudefishing’ > Source: > Published: 2026-07-22 15:01:26+00:00 *Substack has a name for the moment you realise a post you admired for its human voice actually came from a bot: Claudefishing. Now it is handing readers a button to catch it, powered by the AI detector Pangram.* The newsletter platform has launched an AI-detection feature. In a pointed post [titled](https://post.substack.com/p/against-claudefishing) “Against Claudefishing,” chief executive Chris Best said readers can now scan posts, notes, replies, and comments. The tool estimates how much of the text a person wrote, and how much came with AI assistance. The scan works on text longer than 100 words, published from today onward, and the result shows only to the person who asks for it. It is live on the web and iOS, with Android to follow. ## What Claudefishing means Best is careful to say the problem is not AI itself. Substack uses AI to build software and product features, he noted, and plenty of writers lean on it thoughtfully. The problem is the mismatch: readers investing their attention in something with no human thought behind it. “Claude has a lot to offer,” he wrote. “But when I want Claude’s opinion, I’ll ask Claude.” The stakes are rising fast. Pangram [estimates](https://www.pangram.com/blog/ai-in-your-feed) that as much as 40% of text on some social platforms is now AI-generated. Substack, which pitches itself as an economic engine for culture built on trust between people, wants to keep its feed from turning into that kind of [slop](https://thenextweb.com/news/ai-slop-crackdown-faceless-creators-collateral-damage). ## Not a ban The tools cut both ways. Writers can run the [detector](https://thenextweb.com/news/meta-ai-detector-cropped-images-watermark) on their own drafts before publishing, add a “How I make this” statement to explain their process, and report any scan of their work they think is wrong. Substack frames the whole thing around a simple line: people should know what they are getting. It is an unusually direct move in a market drowning in [AI-generated feeds](https://thenextweb.com/news/reddit-ai-marketing-slop-geo-crackdown). The post even took a swipe at LinkedIn, the platform many see as ground zero for machine-written earnestness. ## The catch The snag is that AI [detection](https://thenextweb.com/news/ai-cheating-tools-detectors-students-schools) is not settled science. Pangram is accurate, but not perfect, and The Atlantic has [warned](https://www.theatlantic.com/technology/2026/05/pangram-ai-detection-accuracy/687381/) that over-trusting it risks turning accusations into a witch hunt. Substack itself concedes the tool cannot tell whether real care went into a piece, only whether AI shaped the words. Critics go further. Any detector, the developer Perry Metzger argued, lets you fine-tune an AI to evade it, so such tools cannot work for long. The researcher Mor Naaman called detection a losing battle: if readers scan, writers will scan too, and route around it. As one Substack commenter put it, the presence of AI does not prove the absence of a human. ## A very human business Underneath sits a bet. As Business Insider’s Peter Kafka [put it](https://www.businessinsider.com/substack-anti-ai-slop-plan-label-2026-7), if AI wins, “made by humans” could be a very good business. Substack is wagering that [proof a person did the work](https://thenextweb.com/news/brown-university-ai-cheating-in-person-exam-serrano) becomes the scarce, sellable thing. That the fight is playing out over newsletters, of all places, only shows how far the question now reaches. ## Get the TNW newsletter Get the most important tech news in your inbox each week.