Terms of service

Last updated 1 October 2026

Draft, pending legal review.

HumanizeWord is pre-launch. The operating entity and governing jurisdiction are not yet settled, and both are left explicit below rather than filled in with something untrue. This page accurately describes how the service behaves today; the legal framing around it is not final.

These terms cover your use of HumanizeWord at humanizeword.com. The service rewrites text you give it so that it reads as human writing.

Who you are contracting with

[Operating entity and jurisdiction to be confirmed.] We have deliberately not named one here yet. Until it is settled, treat this document as a description of how we operate rather than as a finished contract.

What we promise, and what we do not

We promise to run the rewriting pipeline on the text you submit and to charge you the usage-based fee described under Billing. That is the whole of it, and the next part matters more:

  • We do not promise any particular result from any AI detector. Detectors are third-party models that change without notice. Our own detector score is one signal from one model, shown so you can judge the output — it is not a prediction about what any other tool will say, and no refund is owed if another tool disagrees with it.
  • We do not promise the output is correct. The pipeline rewrites wording; it does not verify facts. Read what it gives you before you use it.
  • We do not promise a rewrite will pass our own checks. A run that fails validation is still charged, because it consumed the same inference as one that passed. We bill what computation costs, not what it achieves.
  • The service is provided as is, without warranties of merchantability or fitness for a particular purpose, to the extent the law allows.

Billing

  • Each run is charged a usage-based fee, calculated from the computing work that run consumes, with a minimum charge per run. There is no subscription and no per-word rate.
  • You are shown an estimate before a run starts, the amount climbing live while it works, and the actual amount when it finishes. The estimate is an estimate: because the pipeline is an agent loop, the same text can be charged different amounts on different runs — in our own measurements, by up to roughly four times.
  • Balance is prepaid, in USD, funded by crypto deposit. We credit the USD value of what arrives on chain and absorb the payment provider's fee ourselves.
  • Deposits are not refundable to a bank account. We have no fiat payment rail, so there is no mechanism to reverse one. Deposit what you intend to spend.
  • Signup credit expires. Its amount and expiry are shown on your billing page.

Acceptable use

You may not use HumanizeWord to:

  • submit work as your own where the rules that apply to you forbid it — academic submissions, professional certifications, and similar. Whether that applies is your responsibility, not ours, and we have no way to know.
  • produce material that impersonates a real person or organisation;
  • generate content that is unlawful where you are, including material that sexualises minors, incites violence, or facilitates fraud;
  • submit text you have no right to, including other people's copyrighted work or personal data you were not permitted to share;
  • attempt to extract our prompts, circumvent rate limits, or resell access without agreement.

We may suspend an account for a clear breach. Your remaining balance is not forfeited by a suspension, but we may decline to process further runs.

Your text

You keep ownership of what you submit and of what comes back. We need a limited licence to process it — to rewrite it and store it so you can read your own runs — and nothing more. We do not train models on your text. You can delete the text of any run, or of all your runs, from Settings; the billing record stays, because our ledger is append-only and removing it would make your own usage history stop adding up.

Your text leaves our systems

Rewriting sends the text you submit out for processing and returns a rewrite. See the privacy policy.

Labelling

Output is machine-generated and we label it as such internally in every case. We will not describe it as human-written, and nothing in the product certifies it as such.

Liability

To the extent the law allows, our total liability to you is limited to the amount you have paid us in the three months before the claim. We are not liable for consequences of using the output — academic penalties, professional sanctions, lost business, or anything else downstream of a decision you made about text we rewrote.

Changes and ending

We may change these terms; material changes will be announced in the app before they take effect. You can stop using the service at any time. If we discontinue the service we will give notice and a window to withdraw any unspent balance by the means then available.

Contact

[email protected], or [email protected] to report misuse.