MyDetector is an online AI detector that estimates whether a passage was written by a person or generated by a language model, and it is pitched at text from ChatGPT, Gemini, Claude and other models. A single scan returns a percentage score together with highlighted sentences, so a reviewer can see which parts of a draft drove the result instead of receiving a bare yes-or-no label.
The same account also opens an AI image detector, an AI code detector, a plagiarism checker, a grammar checker and an AI humanizer that rewrites machine-like text. Every one of these tools draws on a shared monthly quota, and each tool consumes that quota at its own rate.
MyDetector's public terms name REFAR as the operator of the website. No further company details, such as a registered address or jurisdiction, appeared on the about, terms, privacy or refund pages reviewed for this profile.
Two facts frame everything else on this page. First, the vendor itself says its score is a probability rather than a verdict. Second, the same site sells rewriting pages marketed for getting text past AI detectors, which sits awkwardly next to a product whose main job is catching AI text. Both points are covered in detail below.
The scoring approach is explained only in general terms. The homepage says detection works by training on large sets of human and machine text and then scoring traits such as repeated phrasing, flat vocabulary and steady rhythm; no model card, benchmark set or version history was published on the pages reviewed.
The AI humanizer takes text pasted from ChatGPT, Jasper or Gemini and rewrites it into more natural writing, with tone presets such as Professional, Colloquial, Formal, Business, Creative and Friendly. It runs on a Basic model and a Pro model that are charged at different quota rates (see Pricing).
The site describes this rewriting capability in conflicting ways. The humanizer page says its team focuses on readable, accurate output "not on gaming detectors." A separate Bypass AI page offers to "bypass major AI detectors," and an Undetectable AI page promises human-sounding text that AI detectors cannot recognize. The about page goes further, saying the Humanize button ensures text passes all AI detectors. Readers evaluating MyDetector as a neutral checker should know that the same vendor markets detection evasion.
The homepage lays out a three-step flow for the text detector:
Two practical details sit around those steps. The homepage describes a switch between the fast checker and the deeper model, so the trade-off between speed and detail can be made for each task. And because the detector charges quota per block of characters, re-checking a long document after each round of edits consumes quota every time.
The homepage groups its examples by field. Each one uses the same scan, so the difference lies in what the reviewer does with the score afterwards.
In each of these settings the score works as a screening step, not a final ruling, because the vendor itself describes the output as a probability. For coursework in particular, the gap between a percentage and proof of misconduct matters more than for a marketing draft.
The homepage names several audiences, and the strongest fits are the ones where a fast probability score is enough:
It is a weaker fit in three situations:
MyDetector pricing is quota-based: every plan includes a monthly quota, and each tool deducts quota at its own rate. The pricing page captured on October 6, 2026 listed monthly prices only; no annual price was shown.
| Plan | Listed price | Monthly quota | Positioning |
|---|---|---|---|
| Free | $0/month | 20 quotas | Quick preview of results |
| Basic | $6.99/month | 1,100 quotas | Daily detection and creation |
| Pro | $14.99/month | 3,000 quotas | High-volume workflows |
| Premium | $29.99/month | Unlimited quota | Power users and growing teams |
| Tool | Quota rate on paid plans |
|---|---|
| AI Detector | 2,000 characters per quota |
| AI Humanizer | Basic model 1,000 characters per quota; Pro model 200 characters per quota |
| AI Image Detector | 10 quotas per image |
| AI Code Detector, Grammar Checker, Plagiarism Checker | 500 characters per quota |
Because the detector is charged at 2,000 characters per quota and the Pro humanizer model at 200 characters per quota, the same balance covers different amounts of text depending on which tools it is spent on.
Billing rules stated on the pricing and refund pages:
Several pages on the site disagree with the pricing table, and none of them is clearly marked as outdated:
AI tool directories list several detectors next to MyDetector. Two of them differ in ways that change the choice:
MyDetector's own distinction is breadth on one quota: text, image and code detection, plagiarism and grammar checks, and a humanizer, all billed from the same balance. GPTZero is the closer match for teams that want detection inside documents and email; Decopy AI is closer in scope, since it also pairs detection with humanizing.
No independent accuracy test of MyDetector was found during this review. The comparison articles that did turn up were published by companies that sell competing detectors or humanizers, so they are not treated as neutral evidence here.
There is a Free plan at $0 per month with 20 quotas, and guests get one trial before registering. Paid plans start at $6.99 per month. Some tool pages describe parts of the service as free with no limit, but the pricing table charges quota for those same tools, so the pages are inconsistent.
The pricing table lists 2,000 characters per quota for the AI detector on paid plans. The pricing FAQ, however, says quotas are deducted by word count, so the two descriptions of the unit do not match.
No. The vendor says no AI detector reaches perfect accuracy, that structured human writing can be flagged by mistake, and that the result is a probability to be weighed with human judgement.
The privacy policy says text is processed temporarily and not stored permanently. The terms nonetheless grant MyDetector a license to use submitted content for its services, and a History page keeps previously humanized texts for logged-in users. Whether detector inputs are used for model training is not stated in the privacy policy.
Yes. The refund policy offers a full refund within 14 days of the first purchase or any renewal, without requiring a reason, and premium access ends as soon as the refund is issued.