AI-generated writing now appears in LinkedIn posts, Substack articles, medical papers, and published books. As more readers and platforms recognize the patterns, writers who never use AI are growing concerned that their own work might inadvertently trigger detection tools. "What's interesting is that these models do speak in a specific way, and you develop a sense of it-you develop the smell of the writing," said George Strakhov, founder of the AI automation consultancy move38 and creator of the AI style linter Slop Tells.
Companies like LinkedIn and Substack have introduced moderation efforts to reduce fully AI-generated text. Detection tools such as Pangram and Turnitin help universities and organizations flag AI content in essays and work documents. For writers focused on craft, the problem is different: they want to avoid sounding robotic by accident. Several designers are now building tools that highlight the specific words, phrases, and formatting patterns that give prose an AI signature.
Two tools that flag AI tells in your writing
Strakhov and others are creating interfaces that catalog the quirks of machine-generated text. These tools scan submitted writing and mark phrases that commonly appear in LLM output. Here are two worth knowing.
LLM cliché highlighter
Programmer Simon Willison published the LLM cliché highlighter on his X account on August 28. The tool tracks 38 signs of AI writing, including rhetorical tics like "The punchline is," "Here's the twist," and "Don't take my word for it." Users paste text into a window or submit a link to an entire article. The highlighter flags suspect sentences in light yellow and direct AI tell matches in a darker yellow. A counter at the top of the page displays the total number of flagged sentences, matches, and chain items-patterns like "no X, no Y"-that appear throughout the text.
For writers who want a quick diagnostic before publishing, the cliché highlighter provides an immediate read on whether their prose carries the verbal fingerprints of a language model. It does not make judgments about whether AI was actually used. It simply points to the phrases that have become overrepresented in machine-generated content.
Why this matters for writers
AI detection is no longer just about policing student essays or job applications. Writers who produce original work now face a credibility problem: their sentences can get caught in filters designed to catch chatbots. Tools like the LLM cliché highlighter give writers a way to audit their own prose for unintentional AI patterns. The goal is not to avoid detection software but to preserve a human voice. As Strakhov put it, developing a sense of the AI "smell" is the first step-these tools simply make that smell visible. For writers looking to sharpen their instincts around machine-generated language, resources like AI for Writers and AI for Technical Writers offer structured ways to build that skill.
Your membership also unlocks:
