Grammarly AI detector review: 76% on the first scan, a sign-up wall after it
The page headline says “Ranked #1 in Quality” and its copy says “Achieves 99% detection accuracy”. In this test the tool read our AI-written paragraph as 76% AI, with no sentence-level view, and then blocked every later guest scan with a sign-up box. We could not scan our other two texts.
Visit Grammarly ↗Not an affiliate link: we have no commercial relationship with Grammarly and earn nothing from this click. Disclosure
One quick, single-number check of a paragraph, with no account, if you only need a rough read and one scan.
Main limit
One guest scan. After it, pasting and clicking “Scan for AI” opened a sign-up box and produced no result, so we have no score for a rewrite or a human text.
Not for the first scan. Later scans were blocked by a sign-up box; we entered no email and created no account
Input limit
A counter appeared once we pasted text: “109/1,400 words”. No minimum was stated and none showed up in our test
Time to result
About 2 to 3 seconds in the automated run. The page says “In seconds”
What you get
One headline (“76% of this text appears to be AI-generated”), a gauge, and a split: “Resembles AI text 76%” and “No AI text patterns found 24%”. No sentence highlights
What is held back
The page says that signing up for Grammarly Pro gives the “AI Detector agent”, which shows “exactly which parts may sound AI-generated”. A guest sees only the overall number
The wall
After the first result, clicking into the text box opened a box titled “Review your AI essay with Grammarly for free” with “Sign up with Google” and “Sign up or log in with email”. On two reloaded pages, pasting text and pressing “Scan for AI” opened a sign-up box again and the result stayed at “—%”
What the page claims
The FAQ says: paste your text, click Scan for AI, and “in seconds, you’ll see what percentage of your text appears to be AI-generated.” It does not mention a one-scan limit. A different session or day might behave differently; we did not test that
Text A, our AI-written education paragraph, read 76% AI: “Resembles AI text 76%”, “No AI text patterns found 24%”. On this same text Grammarly gave 100.00% and Sapling 99.3%, and Copyleaks, GPTZero and ZeroGPT all flagged it. Grammarly gave the lowest AI number of the detectors that scanned it, in this one test.
The result was one number. We saw no highlighted sentences and no explanation next to the score. The “which parts” view is described as a Pro feature.
The page’s own copy says “Achieves 99% detection accuracy and ranks #1 on RAID’s independent benchmark”. We did not check that claim, and one paragraph cannot say anything about accuracy.
The wall came right after the first result. Text F (a human passage) and text D (an Undetectable AI rewrite) were accepted into the box, with counters of 145 and 122 words, but no scan ran. That is the comparison detectors are most useful for, and a guest could not do it here in one session.
Text A: “76% of this text appears to be AI-generated”, split 76% / 24%, counter “109/1,400 words”.Open original ↗The sign-up box, over the 76% result, after we clicked back into the text box.Open original ↗
The one text we could scan. Copy it to check another detector.
Text A · AI-written · education109 words
Artificial intelligence has fundamentally transformed the way students approach their academic work. In today's rapidly evolving digital landscape, AI-powered tools offer unprecedented opportunities to enhance learning outcomes and streamline research processes. Furthermore, these technologies enable learners to access personalized feedback, identify knowledge gaps, and develop critical thinking skills more efficiently than ever before. However, it is important to note that the integration of AI in education also presents significant challenges, including concerns about academic integrity and over-reliance on automated assistance. Ultimately, striking the right balance between leveraging AI capabilities and fostering independent thought will be crucial for educators and students alike as they navigate this new era of learning.
On Oct 29, 2026 an automated browser opened Grammarly’s free AI Detector as a guest, pasted text A and ran one scan. It then tried text F and text D, on the same page and on reloaded pages, and stopped at the sign-up box each time. We recorded the output, the limits and the walls; Jamie C. checked the records and wrote this page. Full method.
Not tested
Texts D and F (an AI rewrite and a human passage): the sign-up box came before the scan
Whether the limit resets by day or by session
Anything behind an account: Grammarly Pro, the “AI Detector agent”, plagiarism checking and the other Grammarly tools
Inputs near the 1,400-word limit and non-English text
The 99% accuracy and RAID benchmark claims on the page
One quick, single-number check of a paragraph, with no account, if you only need a rough read and one scan.
Main limit
One guest scan. After it, pasting and clicking “Scan for AI” opened a sign-up box and produced no result, so we have no score for a rewrite or a human text.
Not for the first scan. Later scans were blocked by a sign-up box; we entered no email and created no account
Input limit
A counter appeared once we pasted text: “109/1,400 words”. No minimum was stated and none showed up in our test
Time to result
About 2 to 3 seconds in the automated run. The page says “In seconds”
What you get
One headline (“76% of this text appears to be AI-generated”), a gauge, and a split: “Resembles AI text 76%” and “No AI text patterns found 24%”. No sentence highlights
What is held back
The page says that signing up for Grammarly Pro gives the “AI Detector agent”, which shows “exactly which parts may sound AI-generated”. A guest sees only the overall number
The wall
After the first result, clicking into the text box opened a box titled “Review your AI essay with Grammarly for free” with “Sign up with Google” and “Sign up or log in with email”. On two reloaded pages, pasting text and pressing “Scan for AI” opened a sign-up box again and the result stayed at “—%”
What the page claims
The FAQ says: paste your text, click Scan for AI, and “in seconds, you’ll see what percentage of your text appears to be AI-generated.” It does not mention a one-scan limit. A different session or day might behave differently; we did not test that
Text A, our AI-written education paragraph, read 76% AI: “Resembles AI text 76%”, “No AI text patterns found 24%”. On this same text Grammarly gave 100.00% and Sapling 99.3%, and Copyleaks, GPTZero and ZeroGPT all flagged it. Grammarly gave the lowest AI number of the detectors that scanned it, in this one test.
The result was one number. We saw no highlighted sentences and no explanation next to the score. The “which parts” view is described as a Pro feature.
The page’s own copy says “Achieves 99% detection accuracy and ranks #1 on RAID’s independent benchmark”. We did not check that claim, and one paragraph cannot say anything about accuracy.
The wall came right after the first result. Text F (a human passage) and text D (an Undetectable AI rewrite) were accepted into the box, with counters of 145 and 122 words, but no scan ran. That is the comparison detectors are most useful for, and a guest could not do it here in one session.
Text A: “76% of this text appears to be AI-generated”, split 76% / 24%, counter “109/1,400 words”.Open original ↗The sign-up box, over the 76% result, after we clicked back into the text box.Open original ↗
The one text we could scan. Copy it to check another detector.
Text A · AI-written · education109 words
Artificial intelligence has fundamentally transformed the way students approach their academic work. In today's rapidly evolving digital landscape, AI-powered tools offer unprecedented opportunities to enhance learning outcomes and streamline research processes. Furthermore, these technologies enable learners to access personalized feedback, identify knowledge gaps, and develop critical thinking skills more efficiently than ever before. However, it is important to note that the integration of AI in education also presents significant challenges, including concerns about academic integrity and over-reliance on automated assistance. Ultimately, striking the right balance between leveraging AI capabilities and fostering independent thought will be crucial for educators and students alike as they navigate this new era of learning.
On Oct 29, 2026 an automated browser opened Grammarly’s free AI Detector as a guest, pasted text A and ran one scan. It then tried text F and text D, on the same page and on reloaded pages, and stopped at the sign-up box each time. We recorded the output, the limits and the walls; Jamie C. checked the records and wrote this page. Full method.
Not tested
Texts D and F (an AI rewrite and a human passage): the sign-up box came before the scan
Whether the limit resets by day or by session
Anything behind an account: Grammarly Pro, the “AI Detector agent”, plagiarism checking and the other Grammarly tools
Inputs near the 1,400-word limit and non-English text
The 99% accuracy and RAID benchmark claims on the page