Test log
AI or Not test data
Raw results for the AI or Not review. The verdict is in the review.
- Tested
- 6 November 2026
- Tester
- Miles Foster
- Account
- Free demo account, signed up by RankDude, nothing paid to the brand
- Tests
- 8
- Screenshots
- 7
- Method
- Published methodology
- Environment
- Google Chrome on desktop
Results
| ID | Test | Key measurement | Score |
|---|---|---|---|
| T01 | Accuracy on AI-Written Text | AI passages caught: 5 of 5 | 5/5 |
| T02 | False Positives on Human Text | English human passages flagged as AI: 0 of 5 | 5/5 |
| T03 | Mixed & Lightly-Edited Text | Mixed passages flagged as containing AI: 1 of 2 | 3/5 |
| T04 | Non-English Text | Passages correct: 2 of 4 | 2/5 |
| T05 | Document & File Support | Ways in that worked: 3 of 10 | 1/5 |
| T06 | Score Clarity & Evidence | Points met: 5 of 10 | 2/5 |
| T07 | Scan Speed | Median per scan: 0.5 s | 5/5 |
| T08 | Pricing & Free-Tier Limits | Free text words a month: 1,000,000 | 5/5 |
| Overall (in the review) | 3.6/5 | ||
Accuracy on AI-Written Text
5/5Tested- Method
- Five AI-written passages of 78 to 91 words (remote work, the Industrial Revolution, sleep, a cover letter, electric cars) were pasted once each; the first verdict counted, with 50 percent AI as the line.
- AI passages caught
- 5 of 5
- Lowest AI score
- 87% AI
- Remark
- Scale 5: 47 to 50 of 50 caught. 5 of 5 is 100 percent, so the grade is 5; the sample is five passages, not fifty.
Raw notes
Five AI-written passages of 78 to 91 words (remote work, the Industrial Revolution, sleep, a cover letter, electric cars) were pasted once each; the first verdict counted, with 50 percent AI as the line.
Scores were 100, 87, 100, 95 and 99 percent AI. All five showed the verdict Likely Contains AI.
A sixth and seventh AI passage in German and Spanish scored 97 and 99 percent AI (see the non-English test).
- Five passages from one model, at default settings, not 50 from six models.
False Positives on Human Text
5/5Tested- Method
- Five English passages from books published before 1900 (Pride and Prejudice, Walden, On the Origin of Species, Moby-Dick and the Autobiography of Benjamin Franklin) were pasted once each, so any AI flag would be a false positive.
- English human passages flagged as AI
- 0 of 5
- Highest English human score
- 4% AI
- Remark
- Scale 5: at most 1 of 50 flagged. 0 of 5 flagged is 0 percent. The two non-English false flags are graded under the non-English test, which this test does not include.
Raw notes
Five English passages from books published before 1900 (Pride and Prejudice, Walden, On the Origin of Species, Moby-Dick and the Autobiography of Benjamin Franklin) were pasted once each, so any AI flag would be a false positive.
Scores were 3, 1, 1, 4 and 2 percent AI, all Likely Real.
The two non-English human passages did not do as well: Don Quijote scored 99 percent AI and Les Miserables 85 percent AI. They are counted in the non-English test.
- Old literary English is far from the student and web writing that detectors wrongly accuse; a clean result here is a minimum. Five passages, not fifty.
Mixed & Lightly-Edited Text
3/5Tested- Method
- Two passages each joined a human paragraph (from Great Expectations and A Tale of Two Cities) to an AI paragraph, in both orders, and were scanned once.
- Mixed passages flagged as containing AI
- 1 of 2
- Scores
- 75% and 13% AI
- Remark
- Scale 3: 10 to 13 of 20 flagged. 1 of 2 is 50 percent, the same share as 10 of 20.
Raw notes
Two passages each joined a human paragraph (from Great Expectations and A Tale of Two Cities) to an AI paragraph, in both orders, and were scanned once.
The first scored 75 percent AI, Likely Contains AI, and the shading sat on the AI paragraph. The second scored 13 percent AI, Likely Real, so the AI paragraph passed as human, although it carried a faint shading.
- Two passages, not twenty. The AI-draft-with-human-edits kind was not run.
Non-English Text
2/5Tested- Method
- Four passages were scanned once: an AI passage in Spanish, an AI passage in German, a human Spanish passage (Don Quijote, 1615) and a human French passage (Les Miserables).
- Passages correct
- 2 of 4
- Human passages flagged
- 2 of 2
- Remark
- Scale 2: 10 to 14 of 24 correct, or human passages flagged across languages. Two human passages in two languages were wrongly flagged.
Raw notes
Four passages were scanned once: an AI passage in Spanish, an AI passage in German, a human Spanish passage (Don Quijote, 1615) and a human French passage (Les Miserables).
Both AI passages were caught: 99 and 97 percent AI.
Both human passages were flagged as AI: the Spanish one at 99 percent and the French one at 85 percent, both Likely Contains AI.
- Four passages, three languages. Chinese was not run. Both human texts are old classics, so a modern writer could fare differently.
Document & File Support
1/5Tested- Method
- Each way in was tried on the free plan: paste, a text file, a Word file, a PDF, a batch, a web address, a 2,000-word scan, an extension, the API and an export.
- Ways in that worked
- 3 of 10
- Remark
- Scale 1: 3 or fewer of 10 points.
Raw notes
Each way in was tried on the free plan: paste, a text file, a Word file, a PDF, a batch, a web address, a 2,000-word scan, an extension, the API and an export.
Worked: paste; a 2,008-word scan accepted in one go; an API with up to five tokens and public documentation. That is 3 of 10.
Did not work: the upload box (a .txt and a .docx through Browse and by drop left the editor empty with no message; a PDF is not in its list of formats); no batch, no check by web address, no browser extension and no export of results that I could find.
- The upload failure may be a fault of my browser set-up, but it did not respond to the real file chooser or to a drop. No API call was made.
Score Clarity & Evidence
2/5Tested- Method
- Each of the ten points was checked on the scans already run, with one passage scanned three times for repeatability.
- Points met
- 5 of 10
- Remark
- Scale 2: 4 or 5 of 10 points.
Raw notes
Each of the ten points was checked on the scans already run, with one passage scanned three times for repeatability.
Met: a plain verdict label (Likely Contains AI or Likely Real); a percentage on a gauge; shading on the sentences it flags; the same passage scored 100 percent AI three times in a row; no upsell blocking the result. That is 5 of 10.
Missed: a stated margin, an explanation of why, a share or download option, a scan history in the dashboard, and a stated limit of the detector on the result.
- Checked on the free plan only; the paid plans may add more.
Scan Speed
5/5Tested- Method
- All 19 scans were timed from pressing Scan Text to the verdict on screen, checking twenty times a second.
- Median per scan
- 0.5 s
- Slowest scan
- 1.1 s
- Remark
- Scale 5: a median under 5 seconds.
Raw notes
All 19 scans were timed from pressing Scan Text to the verdict on screen, checking twenty times a second.
Median half a second (0.51 seconds), fastest 0.46, slowest 1.08, which was the 2,008-word text. Every short passage took between 0.46 and 0.54 seconds. No scan failed or queued.
- Free plan, short passages plus one 2,000-word text.
Pricing & Free-Tier Limits
5/5Tested- Method
- The free plan was used for the whole test; the pricing page and the credits page were read on the test date.
- Free text words a month
- 1,000,000
- Cost of 1,000 words on pay as you go
- $0.005
- Remark
- Scale 5: a free tier of 1,000 words a day or more.
Raw notes
The free plan was used for the whole test; the pricing page and the credits page were read on the test date.
Free: $5 in credits, and the pricing page promises 1 million words of AI text detection every month, with an API key included. It needed a Google sign-in and no card. Text costs $5 per million words on pay as you go, so 1,000 words cost half a cent.
Pro is $5 a month or $4.17 a month billed yearly and adds credits for every kind of detection. Business is $100 a month. Enterprise is custom.
- No plan was bought; checkout was not opened. The monthly renewal of the free credits was read, not waited for.
Coverage
Tested
- 19 timed scans: 5 AI passages, 5 human passages from books published before 1900, 2 mixed passages, 4 non-English passages, 2 repeats of one passage and one 2,008-word text
- Time from pressing Scan Text to the verdict on screen, for all 19 scans
- The file upload box with a text file, a Word file and a PDF
- The API page, the credits page and the pricing page read on the test date
Not tested or limited
- The free plan paid for far more than I needed, but I ran 19 short passages of 73 to 120 words, not 50 + 50 + 20 + 20 + 24 full passages
- Resistance to humanisers and AI model coverage were not run: no humaniser account was bought and every AI passage was written by one model
- Every human passage is from an old book, which is easier to tell from AI than a modern essay
- Chinese was left out of the non-English test
- The file upload box did nothing when I gave it a .txt, a .docx and a .pdf: no text appeared and no message was shown. This may be a fault on my side, so file support is graded on what I saw
- I did not create an API token or call the API
- No plan was bought: the Pro and Business plans were read, not run
- No Trustpilot profile for AI or Not was found, so no star reading is given
Still to test
- The other 45 AI and 45 human passages
- Modern human writing, such as student essays
- Resistance to humanisers and paraphrasers
- AI model coverage across six models
- Chinese text
- The API with a real token