Direct answer
No mainstream AI detector is genuinely "unlimited" in the marketing sense — every tool meters something, whether it is words per scan, credits per month, or characters per paste box. What actually varies is the unit of the limit: Turnitin does not charge by word count at all, because institutional licences cover whole submissions, while most consumer "AI detector" sites cap you at a few hundred to a few thousand words per free check and sell top-ups. For a 12,000-word dissertation chapter, that difference decides whether you can check your draft in one pass or have to stitch results together from a dozen truncated pastes [1].
Do AI Detectors Really Have Word Limits, and How Long a Document Can They Scan?
Yes — but the limits are commercial and structural rather than technical. Detection models process text in windows of a few hundred tokens, so length itself is not the constraint; a 25,000-word thesis is not harder to classify than a 2,000-word essay, it is simply more windows. What creates the ceiling is the business model wrapped around the model. Free consumer checkers give away a fixed allowance because inference costs money, and "unlimited" tiers usually hide fair-use clauses or throttle you after a threshold.
Turnitin's AI writing indicator works differently in a way that matters here. The report is expressed as a percentage of qualifying text, and the documentation is explicit that very short submissions may return no indicator at all because there is not enough qualifying text to analyse [2]. That is a minimum-length floor, not a maximum-length cap. Institutional licences are priced per student or per institution rather than per thousand words, which is why a full dissertation submitted through an LMS is assessed in its entirety [1].
The distinction to hold onto: a word cap is a billing artefact, while a minimum-length floor is a statistical necessity. Confusing the two leads students to assume their long document is "too big to check" when in fact the opposite is true — long documents give the detector more qualifying text to work with, and give you a more stable percentage [2].
Why Do Long Documents Get Flagged or Truncated, and How Do Word Caps Affect the AI Score You See?
Truncation is the most damaging side effect of word caps, and it is usually invisible. When a free checker accepts only the first 1,000 words of your chapter, the percentage you receive describes that excerpt — not your document. If your introduction is heavily templated (definitions, scope statements, literature-summary sentences) and your analysis section is written from field notes, the excerpt score can be dramatically higher than the whole-document score. Students then rewrite the wrong section, or panic over a number that was never about their work as a whole.
The percentage itself is also widely misread. Turnitin reports the share of qualifying text flagged as AI-generated, alongside a breakdown of how much of the submission qualified for analysis, so the same absolute amount of flagged prose produces a different percentage depending on document length and how much of it is references, quotations, or tables [3]. A long document with a large bibliography and heavy direct quotation has proportionally less qualifying text, which changes the denominator, not the numerator.
False positives cluster in predictable places: formulaic academic phrasing, standard method descriptions, and prose that closely paraphrases source material. Turnitin's own guidance for educators stresses that the indicator is a signal to start a conversation, not a verdict, and that context — drafting history, version control, cited sources — should carry more weight than the raw figure [3]. That guidance applies just as well to you reading your own report before submission: treat a high segment score as a prompt to look at which paragraphs were flagged, not as proof you have a problem.
How Can You Check a Long Thesis or Dissertation for AI Without Hitting a Word Cap?
The reliable approach is to get the same report your institution would generate, on the complete file, rather than assembling a picture from capped free tools. Turnitin's documentation notes that reports are produced from the submitted document as a whole, with flagged passages highlighted in context, which is exactly the view you need when a 15,000-word chapter is at stake [4]. Checking the full document also removes the excerpt bias described above, so the score you act on is the score that reflects your actual draft.
Practical workflow for long-form work: check the complete file once, read the flagged passages in context, revise only where the flagged prose is genuinely generic or unedited, then re-check the revised version rather than re-checking fragments. Guidance on discussing AI writing with students makes the same point from the other side of the desk — students who see the report before final submission understand the evidence and can explain their drafting process, which is what an academic-integrity conversation actually turns on [4]. Sampling a document in 500-word chunks gives you neither the full picture nor a defensible record.
If your draft is longer than any free tool will accept, that is a signal to use a full-document check rather than to shorten what you paste. Turnitin0 runs complete-document checks on .docx, .pdf, or .txt files between 300 and 30,000 words and under 20 MB, returning the Turnitin AI detection report and the similarity report together, so a dissertation chapter or a full master's thesis fits inside a single order. Files are checked without being added to Turnitin's student paper database, and most orders return in under 15 minutes.
If the word cap is the thing standing between you and a straight answer about your draft, stop splitting your thesis into paste-sized pieces — run the whole file through turnitin0 and read the report your marker would read.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is there an AI detector with truly unlimited words?
No consumer AI detector is unlimited in the literal sense; free tiers cap words or credits, and paid tiers apply fair-use limits. Turnitin does not meter by word count at all, because institutional licences cover whole submissions rather than per-thousand-word usage [1].
What is the maximum word count Turnitin can analyse?
Turnitin processes the full submitted document and reports a percentage of qualifying text, so there is no per-scan word ceiling in the way consumer tools have one. There is, however, a minimum: very short texts may return no indicator because there is insufficient qualifying text to analyse [2].
Why is my AI score higher when I check only part of my essay?
Because the percentage is calculated over the text you actually submitted. A templated introduction checked alone can score far higher than the same introduction inside a 10,000-word document with references and analysis, since the denominator of qualifying text changes [3].
Does a high AI percentage mean I will be penalised?
Not on its own. Turnitin's guidance frames the indicator as a prompt for review, and academic-integrity decisions are meant to weigh drafting history, cited sources, and context alongside the number [3][4].
Can I check a full dissertation before submitting it?
Yes. Turnitin0 accepts .docx, .pdf, or .txt files from 300 to 30,000 words and under 20 MB, returning the AI detection and similarity reports together without adding your file to Turnitin's student paper database.