Fine-print review · documents read November 4, 2026

Undetectable AI review: what the humanizer's terms actually promise

A paid AI humanizer with a free AI detector attached, sold by monthly word allowance to writers, marketers and anyone worried about detector scores.

Zone score3.3/10
Made by
Undetectable Inc.
Paid from
$5/month (10K words, billed $60 yearly)
Free plan
Yes
Terms read
November 4, 2026
Users
3.6/5 (Trustpilot)
Visit Undetectable AI → (affiliate)

Our verdict

The pitch is a money-back guarantee, but the terms narrow it to a refund of credits when output provably scores under 75% human, claimed within 30 days. Everything else is non-refundable, cancelling wipes your remaining words on the spot, and outside California and New York you cancel by email. The privacy policy dates from March 2023 and never says what happens to the text you paste. Buy a small tier, monthly, only if your use is clearly allowed where you will submit the text.

Good fit if you…

  • Marketers rewriting their own AI drafts for tone and readability
  • Users who want a free, no-signup AI detector for quick checks
  • Light users who can live with a small monthly word allowance

Skip it if you…

  • You are a student under a no-AI policy (the vendor itself lists this as misuse)
  • You expect a refund if the tool disappoints you for any reason
  • You need a clear written answer on how your pasted text is stored

The fine print, graded

Five questions, graded A (best for you) to E from Undetectable Inc.'s own pages. How grades work.

C

Pricing clarity · Are prices public and free of traps?

Prices are public and in US dollars, but the page loads on the yearly view with the monthly figure struck through, so the headline numbers are annual commitments. The cheapest tier (10K words at $5/month, charged $60 a year) shows no monthly option, and the monthly price list starts at 20K words for $19. The 50K card shows $42.50 while the slider data lists $42, and its yearly total is not shown. A '6 Months Free' badge sits on that card without saying what it applies to. The page says nothing about sales tax.

Source
D

Cancel & refunds · Can you leave, and get money back?

The terms (updated 01/01/2025) say refunds are not provided unless required by law. The only exception is a refund of credits when a submission 'provably flags as less than 75% human on a reputable AI detector', requested within 30 days. California and New York residents can cancel in the dashboard; everyone else must email contact@undetectable.ai. Cancelling stops future charges but does not refund past periods, and if the company stops operating, yearly plans end without compensation.

Source
D

Limits & credits · What is capped, and do credits expire?

Paid plans are sold as a monthly word allowance. The pricing page reserves 'non-expiring credits' and 'permanent words with no monthly expiration' for Business, which implies personal allowances reset monthly. The terms add that 'If you cancel, you will immediately lose your remaining words', even mid-period. The humanizer page caps a single submission at 10,000 characters, and the free trial at 250 words.

Source
D

Your data · Is your content used to train models?

The privacy policy was last updated 03/18/2023. It covers account and usage data and says information is processed to 'provide, improve, and administer' the service, but it never says whether pasted text is stored, for how long, or whether it trains models. No training opt-out is described. The Ethics page separately says the company used 'the data we've collected over the years' to build deepfake-detection tools. Deletion is by email request.

Source
D

Output rights · Can you use what you make commercially?

The terms never state who owns humanized output. Content you post into interactive features is a 'Contribution', for which you grant a perpetual, irrevocable, transferable worldwide license. The terms also forbid using any output to train or evaluate AI models, and a broadly worded prohibited-activities clause bars using the services or content for 'any revenue-generating endeavor', which sits oddly with a product sold to marketers. Users must be at least 18.

Source

What users say about Undetectable AI

C

User reviews · What do paying users report?

Opinion is split down the middle: Trustpilot shows 3.6 from 1,009 reviews, with 51% five-star and 40% one-star, and it is the only platform with enough reviews to count, so the grade cannot go above B. The most repeated complaint is billing: annual plans charged when people expected monthly, trials that bill right away and renewals with no reminder. The most repeated praise is quick, friendly support that sorts out refunds. Many of the five-star reviews are short and generic, praising support in similar words, which is worth weighing even though Trustpilot shows no review invitations.

PlatformRatingReviewsRead
Trustpilot3.6/51,009November 5, 2026
Product Hunt5/54 (too few to count)November 5, 2026

Recurring complaints

  • Charged for an annual plan when a monthly plan was expected, or billed as soon as a free trial started
  • Auto-renewals with no advance reminder
  • Cancelling takes an email to support, and removing a saved card is hard
  • Refunds that come back partial, with a fee kept
  • Humanized text that still scores very differently from one AI detector to the next

What users like

  • Fast, friendly support by email, with refunds granted in several cases
  • Rewritten text that passes the detectors some reviewers tested it against

Ratings are the platforms' own figures on the date shown. Review sites attract unhappy customers and some vendors invite happy ones, so we average platforms instead of trusting one. How the user grade works.

Undetectable AI plans and prices

PlanPriceBillingWhat you get
Free$0freeFree AI detector with no signup; humanizer trial limited to 250 words; free humanizer uses a basic model
10K words$5/month, billed $60 yearlyannual10,000 words a month, unlimited AI detecting, Human Auto Typer, API compatible
20K words$19/month or $9.50/month billed $114 yearlymonthly or annual20,000 words a month, same feature list
35K words$31/month or $15.75/month billed $189 yearlymonthly or annual35,000 words a month, same feature list
50K words and up$42.50/month or $21.25/month billed yearlymonthly or annual50,000 words a month; a slider runs up to 380,000 words for $209/month on monthly billing
BusinessContact salescustomCustom plans, non-expiring credits, white labeling, priority support, Smart Job Applier

Prices as shown on Undetectable Inc.'s pricing page on November 4, 2026. Taxes may apply.

Pros

  • Prices are listed in US dollars with word counts per tier
  • The AI detector is free and needs no signup
  • Yearly billing roughly halves the per-month price
  • A written credit refund exists when output scores under 75% human
  • The Ethics page states clear do-not-use cases, including cheating in school

Cons

  • No refunds outside the narrow detector-score exception
  • Remaining words are lost immediately when you cancel
  • Most users must cancel by email rather than in the dashboard
  • Privacy policy from 2023 is silent on stored text and training
  • Output ownership is not addressed in the terms
  • Accuracy and bypass claims are the vendor's own and vary from page to page

What Undetectable AI sells

Undetectable AI is two products sharing one brand. The humanizer rewrites AI-generated text so it reads, and scores, more like human writing; it is the part you pay for. The AI detector estimates whether text was machine-written and is offered free without an account. Around them sit add-ons listed in the footer: an essay writer, a paraphraser, a job applier and several image, voice and document detectors.

We read the pricing page, the terms, the privacy policy, the help page, the Ethics page and the two main product pages. We did not run text through the tool, so nothing below is a judgment of output quality.

Read the accuracy claims as marketing

The detector page headlines '99%+ accuracy', then elsewhere cites 95%, 100% and 'up to 100%' depending on the paragraph, and claims 0% false positives. The humanizer page promises text that 'scores as human across all major AI detection tools' and 'up to 100% human-content scores', while admitting the free model 'may be flagged'. None of this is independently verified by us, and the numbers do not agree with each other.

The terms are more modest: the service is provided 'as is' with no warranty on accuracy. That gap matters, because the only refund you can claim depends on proving a low human score on 'a reputable AI detector', a phrase the terms do not define.

The academic-integrity question

Humanizers sit in a contested spot. Many schools and universities forbid submitting AI-generated work, and running it through a rewriter does not change what the rules say. Undetectable's own Ethics page lists 'Cheating in school' and getting around 'no use of AI' policies as unacceptable uses, and says it investigates .edu signups. Yet the humanizer page answers 'Absolutely!' when asked about academic use and pitches protection from 'false flags' by institutional detectors.

Both things can be true: detectors do misfire, and some students worry about false accusations. Our reading is simply this: check your institution's policy first. The terms make you responsible for using the service lawfully, and a high 'human' score is not permission.

Where the money traps are

  • The yearly default. The pricing page opens on yearly prices. The $5 entry tier only exists as a $60 annual charge.
  • Cancel-and-lose. Unused words vanish the moment you cancel, not at the end of the period.
  • Email cancellation. Unless you live in California or New York, the terms route cancellation through an email to support.
  • The guarantee's real shape. It refunds the cost of humanization as credits when output provably scores under 75% human, and you have 30 days to ask.

If you try it, pick the smallest monthly plan, keep screenshots of any failed detector score, and set a reminder before renewal.

Your text and your privacy

For a tool whose whole job is processing your writing, the privacy policy says remarkably little about that writing. It is dated March 2023, talks about personal and usage data, and does not say whether submissions are kept or used to improve models. The humanizer page offers only that data is protected 'with advanced encryption'. If you handle client copy or anything confidential, that silence is reason enough to ask support in writing before you paste.

Two smaller points from the terms are worth knowing. Anything you send support as feedback or a suggestion becomes the company's property outright. And the terms forbid feeding the tool's output into another AI system as training or evaluation data, which matters if you work on models yourself.

Who should pay, and how

The honest use case is narrow: you have your own AI-assisted draft, you are allowed to publish it, and you want it to read less mechanically. For that, a single month on the 20K tier at $19 tells you more than any marketing page, and it limits what you can lose under a no-refund policy. Yearly billing halves the per-month price, but it also locks a year of payments into a contract that, by its own terms, ends without compensation if the company stops operating.

If your goal is to check whether a text looks machine-written, the free detector costs nothing and needs no account. Treat its score the way you would treat any detector's: as an estimate, not proof. And if you are a student, the decisive document is not this vendor's terms but your course policy.

FAQ

Does Undetectable AI give refunds?

Generally no. The terms allow a refund of credits only if a submission provably scores under 75% human on a reputable AI detector, requested within 30 days, or where the law requires a refund.

What happens to my unused words if I cancel?

The terms say you immediately lose your remaining words when you cancel. Only the custom Business plan advertises non-expiring credits.

How do I cancel Undetectable AI?

California and New York residents can cancel in the account dashboard. Other users cancel by emailing contact@undetectable.ai from their account address.

Is the AI detector free?

Yes. The detector page describes it as free with no signup; its input box allows up to 10,000 words. The humanizer's free trial is limited to 250 words.

Can students use Undetectable AI?

The vendor's Ethics page calls cheating in school and evading no-AI policies unacceptable uses, and the terms require users to be 18 or older. Check your institution's rules before using any humanizer on coursework.

Sources

The “Visit Undetectable AI” button is an affiliate link: we may earn a commission if you buy, at no extra cost to you. It does not change the grades. Disclosure.

Alternatives we reviewed