You’ve probably seen the lists: “10 signs of AI writing.” Em dashes, punchy sentences, a clinical tone. And if you’ve ever been accused of writing like a robot when you were just using proper grammar, you know the sting. I have. It feels like being told your voice isn’t yours.
This isn’t a harmless parlor game. The obsession with spotting AI text is creating a new kind of tribalism—where innocent punctuation marks become evidence of guilt, and where real writers start censoring themselves to avoid sounding “too machine.” We are so afraid of sounding like robots that we are policing our own humanity.
Let’s be clear: AI-generated text is everywhere. But the tools designed to catch it are blunt, subjective, and often wrong. The Economist piece on “How to spot AI writing” is a perfect example—it lists traits like “max-impact sentences” and “verbing nouns,” which are also hallmarks of good journalism, copywriting, and even the way we talk in meetings. The em dash is not a wiretap. It’s a punctuation mark.
Worse, this witch hunt is backfiring. Writers are now actively avoiding structures that might trigger suspicion. They’re writing flatter, more generic sentences. They’re stripping out voice. The very thing we fear—a homogenized, soulless internet—is being accelerated by our own paranoia. Obsessing over ‘AI tells’ doesn’t protect us from AI; it just gives us a new way to falsely accuse real writers of being machines.
I’ve seen this firsthand. A friend who writes for a major tech site was told by an editor to “dial back the em dashes” because readers might think it’s AI. She had been using them for years. That’s not quality control—that’s linguistic profiling. And it’s spreading.
So what do we do? First, stop treating detection checklists as gospel. They’re not science—they’re stereotypes. Second, trust the context. If an article is deeply researched, has a unique angle, or makes you feel something, it’s probably human. Neutrality is death in content, but suspicion is death in culture.
We need to choose a side. I’m on the side of assuming good faith until proven otherwise. Let’s focus on celebrating good writing—whether it’s human or AI-assisted—rather than hunting for crime scenes in punctuation. The real enemy isn’t the em dash. It’s the fear that makes us abandon our own voice.
FAQ
Q: Isn't it important to detect AI-generated content to prevent misinformation?
A: Yes, but detection tools are unreliable and often produce false positives. The goal should be context and quality, not hunting for stylistic quirks that humans also use.
Q: So what should writers do if they're worried about being falsely accused?
A: Don't change your natural style. Instead, focus on adding original insights, personal anecdotes, and specific examples—things AI still struggles to generate convincingly.
Q: Isn't the real problem that AI writing is just bad and we should call it out?
A: AI writing is often bland, but attacking its style punishes human writers who happen to write clearly. The bigger threat is the loss of trust in all writing, not the presence of a few em dashes.