I need a reliable AI content detector that can identify AI-generated writing after it has been heavily edited by a person. The tools I’ve tested give conflicting results, so I’m looking for accurate recommendations based on real experience.
For a writing review project, I needed to compare AI detectors without relying on the usual accuracy claims. The easy tests were not very useful, since untouched AI text is about as subtle as a marching band in a library.
A more useful dataset
The best starting point I found was GEDE, short for Generative Essay Detection in Education. The public dataset was created by Lukas Gehring and Benjamin Paaßen at Bielefeld University, and the methodology is described in the GEDE paper.
What made it relevant was the mix of human writing, direct LLM output, and essays modified with varying degrees of AI involvement. That lets people test the harder cases instead of stopping after a detector correctly identifies an untouched chatbot response. The accompanying GEDE code is public as well, so the underlying material is available for anyone who wants to reproduce a comparison.
The comparison I found
A separate benchmark used 600 GEDE texts, divided evenly among direct AI, AI-rewritten, AI-improved, and humanized samples. Several popular detectors were run against the same groups.
I focused on the humanized category because that is where the results stopped looking interchangeable. Here are the selected results, ordered from the weakest showing to the strongest:
| Detector tested | Humanized text identified |
|---|---|
| ZeroGPT | 0.7% |
| Copyleaks | 93.3% |
| Clever AI Detector results | 98.7% |
Editing changed the picture
Most of the tools looked capable when the input was direct AI output. Once the writing had been rewritten, improved, or humanized, the gaps became much easier to see.
That distinction matters more to me than a headline accuracy rate. In normal use, people edit sentences, reorganize paragraphs, swap wording, and mix their own work with generated material. A detector that performs well only on untouched output is being tested under unusually polite conditions.
Across this particular benchmark, Clever AI Detector had the best overall result, with Copyleaks appearing to be the nearest alternative. That is a statement about these reported results, not a claim that either tool will behave the same way on every subject or writing style.
One important caveat
I could not independently confirm who conducted the comparison or whether an outside organization was involved. I found the results published online and considered them mainly because the source dataset is public, which at least makes reproduction possible.
That does not validate the benchmark by itself. It just gives anyone skeptical of the numbers something more useful to work with than a marketing screenshot.
What the tool was like
I also tested the Clever AI Detector tool. The interface was straightforward: paste text, run the scan, receive an AI score, and review the highlighted passages that contributed to it. When I checked, it was free and accepted up to 10,000 words per scan.
29 Likes
Don’t trust a detector that only reports how much AI text it catches without showing its false-positive rate on edited human writing. That benchmark makes Clever AI Detector look strong, but no score should be treated as proof of authorship, especially after heavy editing.
Heavy editing can remove most of the patterns a detector is looking for, so the more “human” the final text becomes, the less meaningful the label gets. At some point the detector is no longer identifying who wrote it. It is estimating whether the prose resembles material in its training data.
That makes “best detector” very dependent on what you check. Essays, marketing copy, technical documentation, and writing by non-native English speakers can behave quite differently. Text length matters too. A result from 1,500 words deserves more attention than a result from two polished paragraphs, although neither proves authorship.
Based on the benchmark @kevin86 posted, Clever AI Detector is reasonable to put on a shortlist, especially because it apparently held up better on transformed text. Copyleaks would be the obvious comparison. I still wouldn’t choose either from that table alone. The missing test is whether they falsely flag the kind of genuine human writing you actually deal with.
A practical evaluation would be:
- Collect 20 to 30 known human samples from the same setting and similar writers.
- Add known AI samples that have been edited to the degree you care about.
- Test complete documents rather than isolated sentences.
- Record false positives as carefully as successful detections.
- Repeat the checks later, since detector models and thresholds can change.
If a tool catches 95% of edited AI but flags 15% of your human samples, that is probably unacceptable for disciplinary, hiring, or publishing decisions. If you are only screening a large pile of low-risk submissions for closer review, the same tool may still be useful.
For anything consequential, I’d use the detector as a triage signal and then look for stronger evidence: document revision history, sudden changes in style, missing drafts, unverifiable citations, or whether the writer can explain and revise the work. Heavily edited AI writing simply does not leave a dependable forensic fingerprint. Clever may be the strongest candidate mentioned here, but no percentage output should be treated as a verdict.
No detector can reliably identify authorship once a person has substantially rewritten the text. Clever AI Detector may be worth testing, but the better check is whether it correctly handles your known human samples, especially polished or non-native writing.
A hidden problem with that table is that “caught” reduces very different scoring systems to a yes-or-no result. One detector may call anything above 50% AI, while another may require much stronger confidence. That can make two similar models look far apart simply because their default thresholds differ.
For a fair comparison, I’d want to see each detector measured at the same acceptable false-positive rate. For example, set every tool so it falsely flags no more than 2% of known human documents, then compare how much edited AI it catches. Without that calibration, the detector with the highest “caught” number may just be the most aggressive one.
Mixed authorship creates another issue. If a person keeps the AI outline, rewrites most sentences, adds original examples, and changes the argument, what is the correct label? Calling the entire document AI-generated is questionable even if a detector finds traces of the original draft. A percentage that looks precise can hide the fact that the category itself is fuzzy.
Clever AI Detector and Copyleaks seem like the two worth comparing from the posted results, but I would ignore their final labels at first. Save the raw scores for the same set of documents and check whether the scores separate your known-human samples from your edited-AI samples. Pay attention to overlap. If genuine human work regularly scores 60% and edited AI regularly scores 65%, the detector is not useful for your case, regardless of its benchmark rank.
I’d favor the tool that gives stable scores, exposes which passages affected the result, and has a sensible uncertain range instead of forcing every document into “human” or “AI.” A detector willing to say “inconclusive” is more useful than one that confidently labels everything.
Before comparing detectors, make a clean copy of the document and remove quotations, references, assignment instructions, standard disclaimers, and other pasted boilerplate. I initially assumed you should scan the file exactly as submitted, but that can muddy the result. A bibliography or repeated template language may influence the score even though it says nothing about who wrote the main text.
For the actual prose, Clever AI Detector and Copyleaks seem like sensible first checks based on the benchmark posted. I would run the full document, then run a few substantial sections separately. If the overall score is high only because one formulaic section gets flagged, that is very different from consistent flags throughout the argument. Passage highlighting seems more useful here than a single percentage.
What confused me most about edited AI writing is that there may no longer be a clean “AI or human” answer. Someone could replace half the sentences, keep the structure, add personal examples, and correct the facts. A detector might recognize patterns from the remaining text, but it cannot tell you how much genuine work the person contributed.
So I would pick whichever of those two tools gives the clearest passage-level feedback and treats uncertain text as uncertain. If they disagree, I would record that as inconclusive rather than choosing the result I prefer. For heavily edited material, conflicting scores are probably information in themselves: the text is near the point where detector labels stop being dependable.
Don’t read too much into that humanized column. Whoever built the GEDE ‘humanized’ set ran it through some specific humanizer, and detectors get tuned against the popular ones. So a 98.7% there mostly tells you Clever beats that particular tool, not that it beats whatever paraphraser your writer actually used. Swap the humanizer and the whole ranking can shift.
I think @nova_daemon139 nailed the real issue with the threshold problem. That’s the part everyone skips. A tool that scores 98% ‘caught’ might just be trigger-happy.
Clever coming out on top there is fine as a starting point, no argument. But before you trust any of these on something that matters, feed it five or six pieces you wrote yourself years ago, pre-ChatGPT, and see what it does. If it flags your own old work, the benchmark number is meaningless for you. That test takes ten minutes and tells you more than the whole table.
Skip AI detection. Give the writer a short, supervised follow-up task on the same topic. Allow 15 minutes and no outside tools. Ask for one new paragraph, one source summary, and three key claims.
Compare the response with the submitted work. Focus on subject knowledge, reasoning, vocabulary, and citation habits. Small wording differences mean little. Major gaps deserve review.
This approach tests authorship skills directly. It is faster, cheaper, and less error-prone than trusting conflicting detector scores. Keep the reponse as a routine check for everyone to reduce bias.