Home / Guides / Guide

How to Spot AI-Written Text

AI writing has tells. Once you see them, you can't unsee them. Here are the seven that give it away, and why detectors alone aren't the answer.

By 📅 Updated ⏱ 4 min read
⚡ Key Takeaways (TL;DR)

Look for em-dashes everywhere, delve, it's-not-X-it's-Y phrasing, uniform paragraph length, and zero opinions. One tell means nothing, four together is a strong signal. Detectors help but formal human writing trips them too. The tells point at blandness more than machines.

Free tool: AI Detector Got suspicious text? Run it through the detector and compare against the tells below. Free, no signup.
Open AI Detector →

The uncomfortable truth

Here's the uncomfortable truth: most AI text gives itself away in the first paragraph, and you've been reading it for years without noticing. Not because you're careless. Because the tells hide in plain sight, in the rhythm of the sentences, the words it can't stop using, the opinions it never has. Once someone points them out, you'll see them everywhere. Sorry in advance.

The seven tells

  • Em-dashes everywhere. Real writers use them sparingly. AI uses them like punctuation confetti, often two or three per paragraph.
  • "Delve." Nobody says delve in real life. AI says it constantly. "Let's delve into" is practically a confession.
  • The it's-not-X-it's-Y construction. "It's not just a tool, it's a revolution." AI reaches for this shape reflexively. Humans rarely do.
  • Every paragraph the same length. Four to five lines, every time, like it was measured with a ruler. Human paragraphs breathe: short, long, fragment.
  • Zero opinions. AI hedges everything and commits to nothing. "On the other hand" appears right on schedule. A human with no opinions is rare. A machine with none is standard.
  • Tapestry, landscape, "in today's fast-paced world." The metaphor drawer it can't stop opening. Real people don't talk like this.
  • Lists of three with identical rhythm. "Faster, smarter, better." "Plan, execute, deliver." Triplets where every item is the same shape.

One tell means nothing, plenty of humans write a stiff paragraph. Three or four together, and you're almost certainly looking at unedited machine output.

Why detectors aren't the answer

AI detectors exist, and they're fine as a first pass. But here's their dirty secret: formal human writing trips them constantly. Corporate memos, academic essays, careful non-native English, all get flagged, because AI was trained to sound exactly like that. And edited AI text slips right through. So use the detector as a signal, then use your eyes. The tells above beat any score.

Why it matters

Unedited AI text is usually unverified, generic, and thought-free. Nobody stood behind it. Nobody checked it. Nobody cared. Spotting the tells isn't about catching cheaters, it's about knowing when you're reading something nobody actually thought about, so you can spend your attention on writing that someone did.

FAQ

What's the fastest way to tell if text is AI-written?

Look for em-dashes in every paragraph, the word delve, it's-not-X-it's-Y phrasing, and every paragraph being exactly the same length. One tell means nothing. Three or four together is a strong signal.

Are AI detectors accurate?

They're decent at flagging obvious AI text and unreliable at the edges. Human writing with formal style gets flagged, and edited AI text slips through. Use them as a signal, not a verdict.

Can human writing be mistaken for AI?

Constantly. Formal, careful, corporate writing trips every detector, because AI was trained to sound exactly like that. The tells point at blandness more than at machines.

Why does it matter if text is AI-written?

Because unedited AI text is usually unverified, generic, and thought-free. Knowing the tells helps you spot content nobody actually stood behind, whether it matters for trust, grading, or just not wasting your time.

Test your eye Paste anything suspicious into the detector, then check it against the seven tells yourself. Free, no signup.
Open AI Detector →