Writers have blamed em dashes for betraying AI authorship online, but the real giveaway lies in a specific rhetorical pattern: negation paired with contrast. When you see lines like “X isn’t just about A —it’s about B,” chances are high that a large language model produced it. The author points out that AI tools default to that “not just… but also” structure to feign depth, then often add quotation marks and triads to tick every stylistic box. Beyond that, two more habits raise red flags. First, the parallel “From… to…” phrasing pops up everywhere—“From budgeting apps to investment portfolios”—even when a human would simply list examples or say “across apps and portfolios.” Second, the “Whether… or…” construction surfaces in sales copy and social captions, as in “Whether you’re an entrepreneur or a business owner,” giving a forced sense of balance. These patterns aren’t mistakes—they’re shortcuts baked into AI training data. You’ll occasionally spot raw ChatGPT artifacts too, like a leftover prompt: “Would you like a concluding paragraph or bulleted summary?” That slip passes the text from machine to reader unedited. None of these signs prove authorship absolutely, but they flag content that likely went through an AI writing assistant without human polish.