Direct answer
Turnitin can accept PowerPoint (.pptx) files as submissions, but its AI writing detector only produces a meaningful AI percentage when the submitted content contains enough extractable prose — Turnitin's own guidance centres the detector on long-form writing of roughly 300 words or more [1]. In practice that means a slide deck with sparse bullets usually returns no AI indicator, while a deck with substantial speaker notes or paragraph-heavy slides may be analysed. Whether detection runs at all also depends on the instructor enabling AI detection for that assignment [3].
Does Turnitin's AI detector actually scan PowerPoint files, or only Word documents?
The short answer is that file-type support and detection support are two different things. Turnitin publishes a defined list of accepted upload formats for assignments, and Microsoft PowerPoint (.pptx) sits on that list alongside Word documents, PDFs, HTML, and plain text files [2]. So a.pptx deck is not rejected at the door — it can be uploaded, processed, and displayed in the Feedback Studio viewer exactly like a Word file [2].
What that list does not promise is that every detection feature applies equally to every format. The accepted-file list is about whether Turnitin can ingest and render the file, which is a technical pipeline question rather than a statement about which analysis engines run on the extracted content [2]. A student who reads "PowerPoint is supported" as "my slides will definitely be AI-scanned" is reading more into the help page than it says.
The practical consequence is that AI detection on a deck is conditional, not automatic. Turnitin's AI writing detection is built around analysing prose, so the amount and type of text inside the presentation matters far more than the file extension [1]. A deck of short bullet points gives the detector very little to work with; a deck where the substance lives in long speaker notes gives it a great deal.
Which file types and submission routes let Turnitin run AI detection on a presentation?
The AI writing report is only generated when a submission clears Turnitin's minimum text requirements; submissions that fall below that threshold simply return no AI indicator rather than a score of zero [3]. That threshold behaviour is the single most important fact for anyone submitting slides, because it explains why two students can upload the same.pptx template and get completely different experiences.
Availability is also gated by instructor configuration. The instructor decides whether AI detection is switched on for a given assignment, so the same file uploaded to two different assignments in the same course can produce an AI report in one case and nothing in the other [3]. File format alone never determines the outcome.
When the report does generate, it does not appear as a loose number floating in isolation. The AI percentage is presented alongside the similarity report, and the underlying writing report highlights the specific segments where the detector believes AI-generated prose is present [3]. That segmentation is what makes the report usable as feedback rather than a blunt verdict.
How can a student see the same AI and similarity report their instructor would get on a slide deck before submitting it?
Turnitin's own framing is that the AI writing report is a conversation aid, not a verdict, and the company recommends that instructors set expectations with students before work is submitted rather than after [4]. That recommendation cuts both ways: it tells students that the report is something they are allowed to understand in advance, and it tells them that the report they eventually see is the same artefact their instructor sees [4].
Because the report mirrors the instructor view, the most direct way to understand your exposure on a deck is to look at an equivalent report on a draft. Institutions frequently build that familiarity through practice or draft submissions, and reviewing how flagged segments are highlighted is far more instructive than reading a policy paragraph [4]. You learn where your speaker notes cross the line, not just that a line exists.
This is also where the 300-word reality of AI detection becomes concrete rather than abstract. If your deck's text is concentrated in notes and paragraph-level content, an equivalent report will show you which passages the detector treats as prose and which it ignores [1]. If your deck is almost entirely short bullets, the exercise mostly confirms that there was little text for the detector to act on in the first place [1].
Reading a policy page tells you that PowerPoint is an accepted upload format; seeing an actual report tells you what your specific deck looks like under the same AI and similarity analysis your instructor will run. turnitin0 lets you upload a draft, including the text-heavy parts of a presentation, and receive the AI detection report and similarity report together so you can act on flagged passages while you still have time to revise.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does Turnitin reject.pptx files outright?
No. PowerPoint (.pptx) appears on Turnitin's list of accepted assignment upload formats, so the file will be ingested and displayed in the viewer [2]. Rejection is not the issue — the question is how much prose the file actually contains for the detector to analyse [1].
Why did my PowerPoint get no AI score at all?
Because the AI writing report is only generated once a submission meets Turnitin's minimum text requirements [3]. A slide deck made mostly of short bullets often falls below that bar, which produces an absent indicator rather than a zero.
Can my instructor turn AI detection off for a presentation assignment?
Yes. AI detection availability is controlled at the assignment level by the instructor, so the same file can yield an AI report in one assignment and nothing in another [3]. Format alone never decides it.
Does the AI report show which slides were flagged?
When a report generates, it highlights the specific segments identified as likely AI-generated text rather than returning a single unexplained number [3]. That is why Turnitin describes the report as a conversation aid for instructors and students [4].
How much text do I need before AI detection is meaningful?
Turnitin's detection model is oriented around long-form prose of roughly 300 words or more [1]. If your substantive content lives in speaker notes or paragraph-style slides, that is the text most likely to be evaluated.