ThesisDraft vs. QuillBot: A Detector Sold Beside a Paraphraser
Check your writing for AI text — free
The first 1,500 words are free, with no sign-up. Every verdict shows how often it is wrong about verified human writing — a figure no other detector publishes.
Need a researched draft to work from instead? From €4.99, anonymous, one-time payment — see pricing.
What QuillBot and ThesisDraft Are
QuillBot is best known as a paraphrasing tool: paste a sentence, get it rewritten in a different register. It has grown into a suite — summariser, grammar checker, citation generator, plagiarism checker — and it also offers a free AI detector.
ThesisDraft’s AI check is a pre-submission check for students, free for the first 1,500 words with no account, €4.99 once for a full document, with the measured false-positive rate printed beside every verdict.
The interesting difference is not feature-level. It is that one of these products sells the check and the rewrite in the same place, and the other deliberately does not.
Side by Side
| Criterion | QuillBot | ThesisDraft AI check |
|---|---|---|
| Primary product | Paraphrasing and writing suite | The check itself |
| AI detector free tier | Yes, account required | 1,500 words, no account |
| Published false-positive rate | Not published | Published per language and length |
| Sells a rewrite of flagged text | Yes, it is the main product | No — polish only, and we say when it will not help |
| Grammar and summarising | Yes | No |
| Tuned for | General writing | Academic prose, EN and DE separately |
| Refuses to give a verdict | No | Yes — German under 400 words |
| Anonymity | Account required | No name, no account, no university |
The Detector Next to the Paraphraser
Put a detector and a rewriter on the same screen and you have built a loop: score the text, rewrite the text, score it again, watch the number fall. It is a satisfying loop. It is also the exact behaviour that university regulations are being rewritten to catch, and the reason paraphrasing tools now appear by name in examination rules across Germany.
We are not going to pretend to be above commercial interest — we sell a €4.99 check. But we do not sell the second half of that loop, and that constraint is written into how this business is run. A check both sides can trust has to give the same verdict to the student before submission and to the examiner afterwards. A tool designed to make a score go down cannot do that, because it only works while the person reading the result does not know it was used.
There is a writing polish here. It removes filler, hedges and tired transitions — what a good writing centre would strike anyway. Where a document is largely machine-written, it tells you that rewording will not fix it instead of selling you the rewrite.
What Paraphrasing Actually Does to a Detector Score
Worth being concrete, because “it lowers the score” is doing a lot of work in most discussions of this.
A paraphraser changes lexical choice and some sentence boundaries. Detectors do respond to that — some of what they measure is surface-level. What a paraphraser does not change is the shape of the argument, the order in which points arrive, which sources are cited where, and whether the reasoning holds together. Those are what a supervisor reads, and what comes up in a viva when someone asks you to explain your own method.
Paraphrasing can also make writing worse in ways that are easy to spot: synonym substitution that breaks technical terminology, and a register that drifts away from the rest of the chapter. Passing a statistical check while reading strangely is not a good trade.
The False-Positive Rate
Measured 9 August 2026 against human documents published before ChatGPT existed:
- English over 400 words — about 3 false positives in 100.
- English under 400 words — about 2 in 100.
- German over 400 words — about 4 in 100.
- German under 400 words — no high-probability verdict is issued at all.
The limits are published alongside: in-sample long-form results, small AI corpora, and a German long-form corpus of Wikipedia prose rather than thesis prose. See the accuracy page for the corpora and the scripts.
What QuillBot Does Better
- Language help on your own sentences. For a non-native writer working through their own argument, a grammar checker and a rephrasing suggestion are legitimate and useful.
- Breadth. Summarising, citation generation, grammar and translation in one place.
- Free access to most of that, which matters on a student budget.
The tool is not the problem. Using it to change how a text scores rather than to say what you meant is the problem, and that distinction is yours to hold — nobody can hold it for you.
The Verdict
Use QuillBot for grammar and phrasing help on sentences you wrote and can defend.
Use our AI check when you want to know what a detector sees, from a tool that publishes how often it is wrong and does not have a rewrite to sell you when the number is high. The first 1,500 words are free, with no account.
Frequently Asked Questions
Does paraphrasing remove AI detection flags?
Sometimes it lowers a score, and that is not the same as fixing anything. A paraphraser changes surface wording while leaving the structure, the argument order and the sourcing untouched, which is what an examiner reads. If a chapter was machine-written, rewording it does not make it your work, and we say so rather than selling you the rewrite.
Is QuillBot's AI detector free?
Yes, it has a free tier, as does ours — 1,500 words with no account. The difference is what comes with the verdict. We publish the measured rate at which our strongest verdict is wrong about verified human writing; QuillBot does not publish one for its detector.
Is using QuillBot on a thesis allowed?
That depends on your university's rules, and many now address paraphrasing tools explicitly in their examination regulations. Light language help on your own sentences is usually fine; running an argument you did not write through a rewriter to change how it scores is the thing regulations are written to catch. Check your own institution's wording — this is not advice we can give for you.
What does ThesisDraft's writing polish do, then?
It removes filler, hedges and tired transitions — what a good writing centre would strike anyway. It does not try to disguise machine-written text, and where a document is largely machine-written we say the rewrite will not fix it.
Check your writing for AI text — free
The first 1,500 words are free, with no sign-up. Every verdict shows how often it is wrong about verified human writing — a figure no other detector publishes.
Need a researched draft to work from instead? From €4.99, anonymous, one-time payment — see pricing.