Make ChatGPT Write Like a Human (Beat Every AI Detector)
This is the exact setup from the video. You teach ChatGPT the tells that expose AI writing, turn them into rules it can never break, and save it to memory once. Every email, essay, and caption after that comes back sounding like you.
What this actually does
AI detectors flag patterns: inflated word choice, identical sentence lengths, robotic transitions, endless hedging. This makes ChatGPT learn those exact patterns from the source, then permanently avoid them. You set it once in Custom Instructions and it applies to every new chat automatically.
Step 1: Make ChatGPT learn the tells
Open a new chat and paste this. It sends ChatGPT to Wikipedia's Signs of AI writing page, the actual list of patterns that expose AI text.
Open the Wikipedia page titled "Signs of AI writing" (from WikiProject AI Cleanup). Read it in full. Then list every single tell that exposes AI-generated writing, grouped by category: word choice, sentence rhythm, punctuation, formatting, and tone. Be exhaustive, include the subtle ones.Step 2: Turn the tells into one permanent rule
Right after it answers, paste this. It compresses everything into a single instruction you can save.
Now turn every one of those tells into a hard rule you must never break. Then compress all of them into ONE permanent instruction, written as direct commands to yourself, that makes everything you write from now on sound naturally human: vary sentence length, use plain words over inflated ones, cut filler transitions, stop over-hedging, no cliches, real specifics over vague summaries. Output only that final instruction, ready to paste, under 1500 characters.Step 3: Save it to memory (this is what makes it permanent)
Copy the instruction ChatGPT just gave you, then:
- Open ChatGPT and go to Settings > Personalization.
- Set your tone to Efficient, it cuts the fluff at the source.
- Open Custom Instructions, and in the box for how ChatGPT should respond, paste the instruction.
- Save.
It is now applied to every new chat automatically. You never paste it again.
Step 4: Test it
Drop in something ChatGPT wrote for you before and have it rewrite under the new rules. Run the result through any AI detector to check.
Rewrite the text below so it reads as fully human under your permanent rules. Keep the meaning identical, change only what trips AI detectors:
[paste your text]Use it right
This is for making your own drafts sound like you actually wrote them. Do not use it to pass off AI work where a teacher, client, or platform requires you to disclose it.
Related guides
Guide
The Research Paper Stack: Four AIs, One Job Each
Perplexity finds the sources, NotebookLM eats the 20 PDFs, Claude attacks your thesis, ChatGPT runs the final pass. AI detectors get skipped entirely.
Guide
Film Once, Post All Week: The Exact Repurpose Stack
One recording becomes clips, 30 captions, an 8-slide carousel, and a 10-tweet thread. One tool per job, the exact prompts included, and the one job you should never hand to AI.
Guide
Keep 4, Cut 4: Eight Chatbots, No Hedging
Eight chatbots, four survive. The verdict word for word, every cut explained, and the 10-tool stack I normally gate behind a DM, printed in full.
Guide
Make ChatGPT Apply to 500 Jobs For You (Harvard Resume Method)
The full system from the video: ChatGPT rebuilds your resume on Harvard's actual guide, then turns on Agent Mode to find matching jobs and apply for you, tailoring each one.
Get the next one first
New prompts every week.
Free. The new drops and the tools behind them, before they hit the feed.
No spam · New issues Sunday · Unsubscribe anytime
Need it custom?
Want this built for you?
Tell me the idea and I’ll build it. An app, a tool, an automation. You don’t need to be technical.
Frequently asked questions
Yes. It works on the free ChatGPT plan. Custom Instructions and Personalization are available to everyone.
No, that is the whole point. Once the instruction is saved in Custom Instructions it applies to every new chat automatically. You set it up once.
It strips out the specific patterns detectors are trained to catch, so outputs come back far more human. No detector is perfect in either direction, so always read it over before you send it.
Yes. Anything ChatGPT writes after you save the instruction, emails, essays, captions and content, all of it.
Making your own writing sound natural is completely fine. Just do not use it to misrepresent authorship where disclosure is required, like graded schoolwork.