The honest answer
is short.

The honest answer is that the tools charging for this were bad, and the good tools cost £15 a month for something that takes three seconds of model time to do.

We built humanise.ai because we'd used the alternatives. They synonym-swapped. They added "moreover" to every paragraph. They billed monthly for a quota you'd hit in an afternoon. And underneath all of that, they positioned the problem as if writing with AI assistance is something to hide, rather than something that almost everyone does now and most institutions are quietly figuring out how to handle.

AI is in your keyboard. It's in your search bar. It's in your email autocomplete. Pretending writers don't use it is already out of date. The useful question isn't "is this person using AI?" — it's "is what they wrote good, accurate, theirs to claim, and appropriate for the context?" A humaniser helps with one piece of that: making sure AI-drafted prose reads the way you actually write, so the thinking — which is the part that's yours — isn't hidden behind AI cadence.

What we think about AI writing

A few things we believe and try to live by:

It's not cheating to think with a tool. Using AI to draft, outline, or restructure is how a lot of people work now. The cheating isn't the tool — it's claiming work as entirely your own when a meaningful part of the reasoning came from somewhere else, or submitting generated prose into contexts that explicitly prohibit it.

Detectors are not truth. They are probabilistic classifiers with meaningful false-positive rates, especially on non-native English writers, formal registers, and edited prose. A detector score is a signal, not a verdict. We build around that reality. See our guide to AI detection.

Humanising is not laundering. Running an essay through a humaniser does not make plagiarised work original, does not make fabricated citations real, and does not make submitting AI-drafted work compliant with a policy that prohibits it. These are three separate problems. We solve one of them — making AI-assisted writing sound the way you write — and we're direct about the other two.

Free should mean free. We pay the model costs. The site supports itself through a partner integration we don't bolt onto the humaniser experience. If we ever add a paid tier, the free product you're using today stays free. That's the commitment.

Who's behind this

humanise.ai is built and maintained by a small team based in the UK. We don't list the full roster publicly because our previous project attracted the kind of mail that makes a public roster a liability, but if you have a press enquiry or a product question that needs a name attached, write to hello@humaniseai.ai and you'll get a reply from a person.

The site runs on Cloudflare Workers, routes through edge locations on five continents, and handles roughly half a million humanisation requests a month as of April 2026. We publish a security page with the full infrastructure and compliance picture.

What we're working on next

• Additional languages (French, Spanish, German — in that order)
• A browser extension that humanises any text field inline
• An API, free for personal use, with sensible rate limits
• Longer-context passes (currently 10,000 characters; we want 25,000)
• Better diff view so you can see exactly what changed and accept rewrites sentence-by-sentence

If one of those matters to you more than the others, email and say so. The roadmap genuinely moves based on what people actually ask for.

How to reach us

• General: hello@humaniseai.ai
• Security: security@humaniseai.ai — or see /.well-known/security.txt
• Press: press@humaniseai.ai
• Policy & abuse: policy@humaniseai.ai

Last reviewed: 2026-04-20.