Six concrete moves
Vary sentence length, cut abstract nouns, add concrete detail, use your own vocabulary, allow stylistic risk, read aloud. Each maps to a specific signal the detector uses.
For students · Rewriting
This is not about gaming a detector. It is about taking prose that reads as generic or synthetic, for whatever reason, and making it sound like the person who wrote it.
Let's be straight about what this page is and isn't. It is not a guide to defeating AI detectors. Detectors improve faster than obfuscation techniques do, and the tactics that "work" this month won't work next month. If your goal is to submit AI-generated text as your own work, this guide will not help you, and the rest of the site is on the teacher's side of that transaction, not yours.
This page is for a different case: your draft, for whatever reason, reads as generic or AI-like, and you want it to sound like you before you submit it. That case includes essays you wrote entirely yourself that flag anyway, drafts where you used AI for brainstorming or grammar and the voice drifted, and essays you've revised so many times the voice is gone.
The rewrite moves above will not reliably clear a modern detector, and the arms race is not winnable at the student-vs-institution scale. The move that works is the one this guide won't script for you: open a new doc, re-read the assignment, and write what you actually think. It is slower. It is also the thing that holds up.
Vary sentence length, cut abstract nouns, add concrete detail, use your own vocabulary, allow stylistic risk, read aloud. Each maps to a specific signal the detector uses.
Manual rewriting that restores voice, not automated obfuscation. We gate our humanizer behind an account and an ethics policy for reasons explained on /humanizer-policy.
These moves reduce stylistic-polish flags. They do not reliably clear detection on AI-generated text, and that is not what the guide is for.
The heatmap tells you whether the moves worked on your essay.
Run a check