Grammarly is a familiar name in writing assistance, offering everything from grammar corrections to plagiarism checks. Its AI detector makes its own promises, advertising 99% detection accuracy and a top ranking on the RAID benchmark. For students already using Grammarly to review their work, we wanted to see how closely its AI detection aligns with Turnitin and how it compares with Academi.cx.
Watch the full comparison on YouTube
How the test worked
Five essays were sourced from IvyPanda, a free essay database with no affiliation to Academi.cx. Selected paragraphs in each essay were rewritten using ChatGPT to introduce AI-generated content. The same version of each document was then submitted to Academi.cx, Grammarly, and Turnitin. We compared how closely Academi.cx and Grammarly matched Turnitin’s overall AI scores and sentence-level highlights.
The results from each detector
- Document 1: Academi.cx detected 56%, predicting 47 to 60%. Turnitin returned 41%, below the range. Grammarly detected 14%, 27 points below Turnitin.
- Document 2: Academi.cx detected 35%, predicting 28 to 39%. Turnitin returned 33%, inside the range. Grammarly detected 15%, 18 points below Turnitin.
- Document 3: Academi.cx detected 38%, predicting 32 to 42%. Turnitin returned 41%, inside the range. Grammarly detected 48%, 7 points above Turnitin.
- Document 4: Academi.cx detected 33%, predicting 26 to 39%. Turnitin returned 27%, inside the range. Grammarly detected 20%, 7 points below Turnitin.
- Document 5: Academi.cx detected 51%, predicting 43 to 55%. Turnitin also returned 51%, matching Academi.cx. Grammarly detected 34%, 17 points below Turnitin.
The one document that missed
Turnitin’s score on the first document came in six points below the lower end of Academi.cx’s predicted range, while the other four documents all landed within the ranges we provided. It was the outlier in this set, and even here, Academi.cx’s sentence-level highlights stayed closer to what Turnitin flagged than Grammarly’s did, which mattered more than the percentage once the reports were placed side by side and the highlighted sentences were compared directly.
Grammarly held its own
Compared to the other detectors tested in this series, Grammarly performed noticeably better, catching real AI generated content in every document and, on parts of document four, actually outperforming Academi.cx on one specific paragraph that Academi.cx only detected in part. That kind of back and forth didn't happen with most of the other competitors tested, where entire documents came back at 0% or missed by 30 or more points. Grammarly's percentage was below Turnitin's on four of the five documents, and its sentence highlights missed full paragraphs, including introductions and conclusions, more often than Academi.cx's did.
The final scorecard
Academi.cx won all five rounds, five to nothing, coming closer to Turnitin's percentage on every document while also matching its sentence-level highlights more consistently. Grammarly's numbers were consistently closer to Turnitin's actual scores than the other detectors tested in this series, and it deserves credit for that. But its highlights still missed key sections document after document, introductions and conclusions especially, which are exactly the sections a student would need to know about before submitting.

