A .pptx file goes in. Something comes back that looks like a confirmation. The assignment page says the submission was processed.
Whether that means anything depends on which of three things actually happened, and only one of them involves AI detection. If you do not know which one ran, you do not know what was checked.
Upload, similarity, and AI detection are separate paths
Turnitin's file requirements page describes what happens when a PowerPoint enters the system. Whether a .pptx is accepted at all depends on how the assignment and integration are configured. If accepted, Turnitin converts the deck into a static PDF. That PDF is what the similarity checker reads.
So far, none of this involves the AI Writing Report.
The AI detection FAQ describes a separate set of requirements. For English-language submissions, the accepted file types for an AI report are .docx, .pdf, .txt, and .rtf. The .pptx extension is not on that list, so a deck uploaded as a PowerPoint file falls outside what the AI report accepts, whatever happened on the similarity side.
So a confirmation that your file was processed could mean a similarity report exists. It could also mean the file was accepted and no AI report was ever attempted.
Neither case changes the assignment's rules. A missing AI report is not a clearance on AI use: whatever your course says about AI still applies, and your instructor still reads and marks the deck.
What the static PDF conversion removes
When Turnitin processes a PowerPoint, it does not analyze the live deck. According to the file requirements page, the conversion to static PDF keeps text and images but removes presenter notes, videos, and animations. Text with visual effects may not be readable after conversion.
This matters for two reasons. First, if your presenter notes contain substantive content that an instructor expects to review, that content does not make it into the similarity check from a direct .pptx upload. Second, the readable text on slides after conversion is whatever survived as plain, static characters. If you relied on animation to reveal text progressively, the converted version may present it differently than you intended.
The 300-word threshold is about prose, not slide count
Even if you export the deck to PDF yourself and submit that, the file-type gate is only one requirement. The FAQ states that for English-language submissions, AI detection analyzes qualifying text, which it defines as prose sentences in long-form writing. Lists, bullet points, and short non-sentence structures are excluded from analysis.
Slides tend to be heavy on exactly those excluded forms. Headlines are fragments. Bullet points are often phrases, not sentences. Labels, captions, and chart annotations are not standard grammatical prose.
The FAQ also sets a minimum of 300 words of qualifying prose for an AI report to generate. If the prose sentences across an entire slide deck add up to fewer than 300 words, no AI report generates at all. This is separate from the Similarity Report's own 20-word minimum for extractable text, which is about whether the pipeline can read the file.
Do not pad slides with unrelated prose to cross the 300-word threshold. Adding text that does not belong on a slide in order to trigger or influence a detection report is bad practice.
Three checks before you interpret a result
Before reading anything into a Turnitin outcome on a slide deck, ask three questions.
Which report was generated? A Similarity Report shows text matches against sources. An AI Writing Report gives the share of qualifying prose that the detector judges likely to be generated or modified by AI. The two are calculated independently, and having one tells you nothing about whether the other exists.
What file was actually analyzed? A direct .pptx submission becomes a static PDF with notes and animations stripped. A separately exported PDF may preserve text differently. The format of the file that entered the pipeline determines what the pipeline could see.
How much qualifying prose is in the document? The AI percentage, when it exists, reflects prose sentences only. It is not a percentage of all text on all slides. A deck with 800 words total may have far fewer than 300 words of qualifying prose, which means no AI report generates.
What to keep
If you may need to show how a deck came together, hold on to your working files: the outline, your notes on where material came from, and saved drafts with dates on them. A version history is a more direct record than a detector percentage. And if presenter notes matter for grading, confirm with the instructor how they should be submitted, because a direct .pptx upload will not carry them through.
Top comments (0)