Our ethics

Last updated: August 30, 2026

The AI-humanizer category is ethically messy, and a lot of tools in it lean into that. We have chosen a narrower, more honest position, and we want it on the record so you know what you are using.

What HumaText is

HumaText is a writing-polish tool. It takes a draft — often one that was AI-assisted — and makes it read more naturally: tighter phrasing, varied sentence rhythm, fewer of the stock words that make text feel machine-made. It is meant to get a model’s voice out of the way of yours, the same way a grammar checker or a human editor would.

What HumaText is not

• It is not a ghostwriter. It does not write your essay, email or paper for you. You bring the draft; it polishes.
• It is instructed not to invent or alter facts, names, numbers, dates, quotations, or citations, but automated rewrites can still make mistakes and require source review.
• It is not a guaranteed way to “beat” any AI detector. We make no such promise, because no honest tool can — detectors change constantly and disagree with each other.
• It is not a license to misrepresent authorship where the rules you are under prohibit it.

Where we draw the line on framing

We’re honest about what HumaText is: it rewrites your draft to read more naturally — varied sentences, direct phrasing, your meaning kept. What we don’t do is promise a “guaranteed undetectable” result, or that any specific detector will pass it — no honest tool can, because detectors change constantly and disagree with each other. Use it on writing you own or are permitted to revise, and follow the AI policy of any school, employer, journal, or platform you submit to.

Responsible use is on you — and we help

You are responsible for following the rules of any school, employer, journal or platform you submit to. Many institutions allow AI-assisted editing of your own work; some do not; some require disclosure. HumaText is a tool, and like any tool it can be used well or badly. We build it to support the good uses — clarity for non-native English writers, tightening your own rough draft, making professional writing read less stiff — and we say plainly that using it to pass off unoriginal work as your own, against your institution’s rules, is not what it is for.

On detection scores

Our writing-pattern check and any score we show are internal signals, not verdicts. AI detectors are imperfect and regularly flag genuine human writing, especially from non-native English writers. Treat any score as one weak signal about writing quality, never as proof of authorship.

Your text and your privacy

We do not use your input to train any AI model. For details on what we store and for how long, see our Privacy Policy. For how we secure it, see our Security page.

Questions

If our positioning ever feels out of step with how the product behaves, tell us: support@humatext.com. We would rather hear it.