The mechanical tells (fixable automatically)
1. Em-dash density
The most famous one. Models produce em dashes at several times the human rate. One or two per page is normal writing; one per paragraph is a pattern. Our em-dash remover rewrites them.
2. Smart quotes, ellipsis characters, markdown crumbs
Curly quotes where straight ones belong, the single … character, stray **asterisks** and ### header marks from copied chat output, and emoji bullets (✅ 🚀) all mark text as pasted from a chatbot.
3. Hidden characters
Zero-width spaces, word joiners, and narrow no-break spaces frequently ride along in AI output. Invisible to you, obvious to software. See the hidden character remover.
The phrasing tells (you have to edit these)
4. "It's not just X, it's Y"
The contrast-reframe is chatbot rhetoric's favorite move. Once you see it, you can't unsee it.
5. Chatbot vocabulary
Delve, tapestry, testament to, seamless, robust, leverage, elevate, multifaceted, pivotal, game-changer, unlock the power of, vibrant, bustling. Each word is legitimate; the cluster is a fingerprint.
6. Throat-clearing openers
"In today's fast-paced digital landscape…", "It's important to note that…", "When it comes to…" — filler that postpones the point. Humans in a hurry just say the thing.
7. Formula conclusions & connector pileups
"In conclusion…", "Ultimately…", and paragraphs stitched with Moreover / Furthermore / Additionally. Structured like an essay template, because it is one.
8. "Whether you're a X or a Y"
The audience-flattering construction ("Whether you're a startup founder or a Fortune 500 CEO…") appears in AI marketing copy at extraordinary rates.
How to use this knowledge
If you write with AI assistance — which is normal now — run your draft through the checker above. Fix the mechanical artifacts automatically, then rewrite the flagged phrasing in your own voice. The result reads like you, because after that editing pass, it is you. That's the honest version of "humanizing" AI text: actual editing, not detector tricks.