Editorial standards
how we research and correct.

Who writes humanise.ai

Articles are written and edited by the humanise.ai editorial team, a small group based in the UK. Pieces are published under the shared byline "humanise.ai Editorial" and every claim is checked against its source before publication. If you need a named contact for a press or research enquiry, write to hello@humaniseai.ai and a person will reply.

How we research

We work from primary sources first. In order of preference, that means:

1. Peer-reviewed research and preprints, such as the detector studies we cite throughout the site, linked to the paper itself rather than to a news story about it.
2. The vendor's own pages for prices, limits and product claims, labelled as the vendor's claim, with the date we checked them.
3. University and sector guidance for questions about academic policy, linked to the institution's own page.
4. Reputable reporting only where no primary source exists, named in the sentence that uses it.

We do not cite forum posts, anonymous screenshots or other roundups as evidence for a number. When a source cannot be opened or does not say what a summary claims, we leave the claim out.

How we handle numbers

• One number per fact. A price, limit or accuracy figure is the same on every page of the site. When a vendor changes a price, we update it everywhere and add the date.
• Every figure is sourced in the same sentence. You should never have to hunt for where a number came from.
• Vendor claims are labelled. "GPTZero says" is different from "an independent study found", and we write it that way.
• Dated checks. Prices and plan limits carry a "checked" month, because they change often.

What we test and what we do not

Some older articles describe hands-on checks of specific tools. Newer research pages state their method plainly: where we compiled vendor information and published studies without running our own accuracy tests, the page says so in a method note. We do not invent scores, and we do not present a vendor's marketing figure as an independent result.

Dates and updates

Every article shows an "Updated" date that matches the date in its structured data. We move that date only when the content changes, not to look fresh. When a correction changes what an article says, we note it on the page.

Corrections

If you spot an outdated price, a broken source link or a claim you think is wrong, email hello@humaniseai.ai with the page address and what you think should change. We check every correction request against the original source. When we change a fact, we update the page and its date; when a correction changes the meaning of an article, we add a short note saying what changed.

Independence and interests

We run a free AI humaniser, so we have an interest in this market, and we say so where it matters. We review detectors and other humanisers on the same fields: what is free, the limits, what the tool reports, its main limitation and the best independent evidence available. Where a paid tool is a better fit for long documents, teams or an API, we say that too, and we point readers to the option we think fits. Product names belong to their owners, and nothing here is a claim of partnership with any vendor named on the site.

Illustrative examples

Some articles include before and after passages written to show an editing technique. These are labelled as illustrative, so they are never mistaken for test results or real student work. Every published claim is checked against its source before the page goes live, and our AI policy covers how the humaniser itself handles your text.

What we cover

We cover three things: how AI writing differs from human writing and how to edit it, how AI detectors work and where they fail, and the practical questions students, teachers and writers ask about both. Detector reviews follow the same structure on every page: what is free, the limits, what the tool reports, its main limitation and the best independent evidence we can find. If a question you need answered is missing, tell us at hello@humaniseai.ai.

How to cite humanise.ai

If you quote a figure from one of our articles, please cite the original source we link to, not this site, because the source is where the evidence lives. If you want to cite one of our explainers as a whole, use the article title, "humanise.ai Editorial" as the author, the "Updated" date shown on the page and the page address. Researchers and journalists who need clarification on how a page was compiled can email hello@humaniseai.ai, and we will explain which sources were used and when they were checked.

Responsible use

Our guides treat humanising as editing your own draft. We do not publish advice on deceiving teachers, employers or institutions, and our articles on detection point readers to their institution's policy. For the evidence on what detectors can and cannot show, see how AI detectors work and whether humanised text can still be detected.

---

Last reviewed: 2026-10-22.