To humanize AI text, rewrite it so it has a person's habits instead of a machine's: plain words, sentences of different lengths, the specific detail, and no wrapper around the message. There are two quick ways. Paste the text into Hemmingway's free AI humanizer, or skip the rewrite and write it with Hemmingway-1.5 from the start.
Hemmingway-1.5 is our undetectable AI. In our test, 90% of what it wrote passed GPTZero and four other top AI detectors on the first try, and by its Hugging Face card, 89% of its rewrites of AI text passed the same detectors. Real human writing passed 86%.
Way 1: humanize text you already have
- Copy the AI text. Leave out the chatbot's own lines around it, such as "Certainly! Here's a draft" and "Let me know if you'd like any changes".
- Paste it into the AI humanizer. It rewrites it on the page, with no account.
- Say who will read it. A teacher, a client, a friend. The same facts should not sound the same to each of them.
- Say how it should sound. "Plain and direct" or "warm but short" is enough.
- Add one thing only you know. A name, a day, a number. Specific details are what AI text leaves out.
- Use it. The rewrite keeps everything your draft said, in the same order, without adding facts that were not there.
Way 2: write it with Hemmingway from the start
The other way is to never have AI text to fix. Ask Hemmingway for the email, the essay or the post, and Hemmingway-1.5 writes it like a person in the first place: just the text, in the register the reader needs, with nothing around it. There is no second step.
The free tools do this for each kind of writing. The AI essay writer writes a whole essay from your question and points, the AI email writer writes the email, and the paragraph rewriter takes one paragraph at a time.
Before and after
A stiff email, for a client, rewritten in a plain, direct tone. A rewrite along these lines is what to expect:
Before: I hope this message finds you well. I wanted to reach out regarding the upcoming project timeline and kindly request an update at your earliest convenience.
After: Hi, could you send me an update on the project timeline when you get a chance?
The opening of an essay paragraph, in a plainer voice:
Before: Social media has had a profound impact on modern society, transforming the way people communicate, interact, and share information.
After: Social media changed how people talk to each other, and how news gets around.
A line of advice, without the stock words:
Before: In today's fast-paced world, effective time management is crucial. Moreover, it is worth noting that prioritizing tasks can significantly enhance productivity.
After: Managing your time matters. Do the most important work first and you get more done.
Each one says the same thing as before. What went is the filler, the stock words and the even rhythm.
What AI detectors look for
Detectors read patterns in the text. What gives AI writing away is not one word but its habits:
- Even rhythm. Sentences of the same length, paragraphs of the same size.
- Stock words. "Moreover", "crucial", "delve", "it's worth noting", "I hope this finds you well".
- The wrapper. An opening that restates the ask, and a closing that sums up or offers more help.
- Everything in threes. Three adjectives, three reasons, three bullet points, whatever the content needs.
- Nothing specific. General lines where a person would give the name, the day or the number.
Swapping words for synonyms leaves all of that in place. A humanizer has to change the habits, and a writer that never had them skips the rewrite.
What our test found
| Text | Passes GPTZero and four other detectors |
|---|---|
| Written by Hemmingway-1.5 | 90% |
| AI text rewritten by Hemmingway-1.5 | 89% |
| Written by people | 86% |
| Answers from six other big models | 13% to 26% |
The detectors are GPTZero, Fakespot from Mozilla, Vanguard, WasItAIGenerated and MELD. A text passed when they read it as human with at least 90% confidence, and each text went through once, with no retries. The 89% for rewrites is from the model's Hugging Face card. The rest is in our AI detector test, and does AI writing pass GPTZero? has GPTZero on its own.
Humanizing by hand
You can do the rewrite yourself. Cut the first and last paragraph, replace each general line with the specific fact behind it, let one sentence be four words long, drop the stock words, and read it aloud. How to make AI text sound human goes through each habit and what to change. For a side-by-side of the humanizers you can use, see the best AI humanizers.
Try HemmingwayDownload the app
Common questions
What does it mean to humanize AI text?
It means rewriting text an AI wrote so it reads like a person wrote it, to people and to AI detectors. The meaning stays. The habits that give AI away, such as the even rhythm, the stock words and the wrapper, go.
How do I humanize AI text for free?
Paste it into Hemmingway's free AI humanizer. It rewrites it on the page with no account and no card. Hemmingway itself is free to try, with no card needed.
Does humanized text pass GPTZero?
Hemmingway's does, nearly nine times in ten. By its Hugging Face card, 89% of Hemmingway-1.5's rewrites of AI text passed GPTZero and four other top AI detectors.
Is it better to humanize AI text or write it with a human-like AI?
Writing it with a human-like AI saves a step. In our test, 90% of what Hemmingway-1.5 wrote from scratch passed GPTZero and four other detectors, and 89% of its rewrites of AI text did. AI humanizer or AI writer compares the two.
Can I humanize text from ChatGPT?
Yes. Paste ChatGPT's answer into the AI humanizer, without its opening and closing lines, and it rewrites it in a person's voice. How to make ChatGPT undetectable goes through it with an example.