I taught ChatGPT to stop writing like ChatGPT. Permanently. You know the tell. T
By Joanna Lambadjieva · October 8, 2026 · Curated by George's Blog
I taught ChatGPT to stop writing like ChatGPT.
Permanently.
You know the tell.
The em dash in every sentence.
"In today's fast-paced landscape…"
"Let's dive in and unlock…"
"This isn't just X, it's Y."
The second your reader sees it, they know.
A human didn't write this.
And they trust it less.
The problem was never ChatGPT or Claude
It was that nobody told it what to stop doing.
So I built the instructions that strip the slop out:
✅ Kills em dashes and semicolons on sight
✅ Bans the giveaway words: delve, unlock, tapestry, game-changer, realm, harness
✅ Cuts the filler: just, really, actually, very, basically
✅ No fake hype, no "in conclusion," no clichés
✅ Short sentences. Active voice. Talks to "you," not at everyone.
Here's the part that changes everything.
You don't paste this every time.
You drop the compact version into Customize ChatGPT/Claude once.
After that, every new chat writes clean by default.
Set it once. Forget it forever.
And to be clear about what this does.
It doesn't write your ideas for you.
It stops the writing from announcing it came from a robot.
The voice, the judgment, the message stay yours.
This makes your AI drafts sound like you, faster.
I've put the fast version into the checklist below 👇
And if you want the full setup, I've built a step-by-step Notion board.
Both prompts (full and compact),
the exact words to ban,
how to save it as your default,
and the troubleshooting for when it drifts.
Just let me know in the comments and I'll send you the link.
💡 PRO TIP: Save this post so the checklist is there before your next draft goes out.