Grubby AI detector review

Grubby AI publishes no AI detector of its own: its sitemap listed 44 public pages on 9 November 2026, none of them a checker, and grubby.ai/ai-detector redirected to the login screen the same day. People searching for a Grubby AI detector usually mean one of two things, a checker inside the product or the detectors the humanizer claims to pass, and this review answers both from what the vendor publishes. The humanizer itself is reviewed in the Grubby AI review, and vendors that do ship a checker are listed in the humanizer and detector list.

Is there a Grubby AI detector at all?

Not one the public can see. The sitemap read on 9 November 2026 lists the homepage, pricing, the guarantee page, login and sign-up, 32 blog posts and five service pages, each titled as a way to get past a named detector. No page is a checker, and the address a user would guess, grubby.ai/ai-detector, sends a signed-out visitor to the login page. Whether a scan feature sits behind the login is not stated anywhere on the public site, so this review does not claim one exists or that one does not.

What the site does publish is a claim about other people's detectors. The homepage lists GPTZero, Originality.ai, ZeroGPT, Turnitin, Winston AI, Content at Scale and Copyleaks and states that its humanizer achieves 99 percent or higher human scores on them, read 9 November 2026 (Grubby AI pricing page carries the same positioning). No input set, no date and no method accompany the figure.

What does the vendor publish about Turnitin?

A refund promise and one number. The guarantee page states that if text humanized in Grubby's Academic mode is flagged as AI by Turnitin, the vendor refunds the cost of the subscription, and it states a 99.8 percent pass rate on Turnitin's AI detection across thousands of its own Academic mode submissions, read 9 November 2026. The same page says about 14,000 students use the tool each week. None of those figures carries a dataset, a date or a third party.

Turnitin's detector is not available to the public, so this site reports only what Turnitin publishes and what peer-reviewed studies found, never a Turnitin pass or fail. A vendor that states a Turnitin pass rate is reporting what its own customers told it, or what it saw in accounts it does not describe. Either way the reader cannot check it.

What does the research say about pass-rate claims?

The published research explains why a paraphrase moves detector readings and why a vendor's pass rate is not the end of the matter. Weber-Wulff and colleagues tested 14 AI text detectors and found every one below 80 percent on their accuracy measure, with accuracy falling further on paraphrased text (Weber-Wulff et al. 2023, International Journal for Educational Integrity, read 9 November 2026). Krishna and colleagues showed a paraphrase model dropping DetectGPT's detection rate from 70.3 percent to 4.6 percent at a 1 percent false positive rate (Krishna et al. 2023, arXiv, read 9 November 2026).

Detector makers answer by training against paraphrase. GPTZero's technology page describes a "Paraphraser Shield" against paraphrased text and states a false positive rate of no more than 1 percent on its own benchmark (GPTZero, How GPTZero works, read 9 November 2026). A pass rate measured in one month can fall the next when a detector updates, and a reading is true for its date. No detector score should be the only evidence in a misconduct case.

What does Grubby AI report?

Nothing from a detector, because it has none in public. What a Grubby customer gets is the rewritten text and, on paid plans, the refund promise. The homepage shows a screenshot it labels as a GPTZero result on a Grubby sample, read 9 November 2026; a vendor's screenshot of a detector reading on a sample the vendor chose is the weakest form of evidence a reader can be offered, since neither the input nor the date can be checked.