Turnitin guide
Can Turnitin Detect AI-Generated Images? What It Checks, and What It Doesn't
Turnitin's AI report reads qualifying prose only, so images go unscored. What the rules say about AI images in coursework, and how teachers can check one.
On this page
No — Turnitin's AI detection reads writing, not pictures. Its AI writing report scores qualifying long-form prose. A photo, illustration or diagram in your file is not something it judges as AI-generated or not. That doesn't make AI images invisible in coursework: teachers have other ways to check them, and most assessment rules already cover them.
The whole page in one table:
| Question | Answer |
|---|---|
| Does Turnitin's AI report check images? | No — it scores qualifying prose only |
| Does Turnitin check images for plagiarism? | Its similarity check compares text, not pictures |
| Does Turnitin have any image checking? | For publishers: PubShield, with Proofig, screens research manuscripts — not student coursework |
| Do the rules cover AI images? | Usually yes: AI use must be acknowledged, whatever the output |
| Can a teacher still check an image? | Yes — provenance, drafts and source files, reverse search, and an AI image detector as one signal |
| Is an AI image detector result proof? | No — detectors make mistakes in both directions |
What Turnitin's AI report actually reads
Turnitin's AI writing detection works on text. It splits a submission into segments of prose, scores each for AI-like writing patterns, and reports the share of qualifying text that looks AI-generated. Only long-form prose in supported languages qualifies: bullet lists, tables, code, references and very short passages are excluded from the calculation. We cover that in detail in how to read your Turnitin AI report.
Pictures are not prose, so they sit outside that calculation entirely. If you paste an AI-generated illustration into an essay, the AI percentage is about the essay's words — not the illustration. The same goes for a poster, a design portfolio or a lab report with generated figures: whatever Turnitin says about the writing, it has said nothing about the images.
For the full picture of the text side — the 300-word minimum, the asterisk for low scores and who can see the number — see does Turnitin detect AI?
Try the text in context
Run a free AI-writing signal check
Paste 300 characters to 350 words. The sample is analysed for a first-pass signal and then discarded.
Does Turnitin check images at all?
Not in the way students worry about. Turnitin's similarity (plagiarism) check compares the text of a submission with its databases and the web; it is not an image-matching tool.
Turnitin does have an image product, but not for coursework. In September 2025 it announced PubShield with Proofig, which pairs Turnitin's iThenticate text screening with Proofig's image integrity checks for research manuscripts — duplication, manipulation and likely AI-generated images in things like microscopy and western blots. It is sold to publishers and research institutions for pre-publication checks, not to schools marking essays.
So for a typical school, college or university assignment, no Turnitin report will tell a marker whether a picture was made with Midjourney, DALL·E or Gemini.
What the rules say about AI images in coursework
Most assessment rules are written about AI use, not about a particular kind of output. In the UK, the JCQ's guidance on AI in assessments lists image generators such as Midjourney, Stable Diffusion and DALL·E alongside chatbots, and says that where AI-generated content is used, the reference must name the AI tool and should give the date the content was generated. Universities' policies usually take the same line: using AI may be allowed, hiding it is not.
That matters more than any detector. If your course allows AI images with acknowledgement, a correctly credited image is fine even though no tool "detected" it. If your course bans AI images, an uncredited one breaks the rule whether or not anything flags it. Check your subject specification and your institution's policy — for UK universities, our AI-detection policy tracker links to each one's current rules.
How teachers can check an image
Without a Turnitin signal, the most reliable checks are about where the image came from, not what a model thinks of its pixels:
- Ask for the working. Sketches, drafts, source files, layers or the original photo with its camera metadata show a process an AI image usually lacks.
- Look for Content Credentials. Images from Adobe tools, some cameras and several AI generators can carry C2PA Content Credentials or an AI watermark that says how they were made. Social networks and messaging apps often strip them, so their absence proves nothing.
- Reverse image search. A picture found elsewhere online is a sourcing question before it's an AI question.
- Use an AI image detector as one signal. A detector rates how generator-like an image looks, and a forensic report can look for signs of editing. Treat it as a reason to ask questions, not a verdict: in NewsGuard's May 2026 test of five popular detectors, real news photos were wrongly flagged as AI about 13% of the time, and small local edits are the hardest case for every tool.
Genutext's AI image detector gives a free AI-generation signal and an optional forensic report, and deletes the image straight after the check. Like every detector, it is a signal for review, never proof.
If you're a student
- Follow your course's rule and credit AI images the way it asks — tool name and date is the common minimum.
- Keep your working files. Drafts, source photos and exported layers are the strongest evidence that an image is yours.
- Don't rely on a detector to clear you. A "likely real" result from any tool is not proof either way; your process and your acknowledgements are what count.
- If you're accused over an image, the same principles as for writing apply: ask what the evidence is and show your process. See how to prove you didn't use AI.
Frequently asked questions
Can Turnitin detect AI-generated images?
No. Turnitin's AI writing detection scores qualifying prose, so images in a submission are not judged as AI-generated or not.
Does Turnitin check images for plagiarism?
Its similarity check compares the text of a submission with its databases and the web. It is not an image-matching tool for coursework.
Does Turnitin have any AI image detection?
Only for publishers: PubShield, announced with Proofig in September 2025, screens images in research manuscripts. It is not part of the coursework checks schools and universities run.
Do I have to credit AI-generated images?
Usually, yes. UK JCQ guidance and most university policies require AI use to be acknowledged, with the tool named and the date given. Check your course's own rules.
How can a teacher tell if an image is AI-generated?
By checking its origin — drafts and source files, Content Credentials, reverse image search — and, as one signal among several, an AI image detector. No detector result is proof on its own.
Sources and further reading
- Turnitin: AI writing detection model
- Turnitin: File requirements for an AI writing report
- Turnitin: Understanding the similarity score
- Proofig: PubShield unites text and image integrity checks in one workflow (22 September 2025)
- JCQ: AI Use in Assessments (April 2025)
- NewsGuard: AI image detection tools often declare authentic content fake (May 2026)
- Genutext: UK University AI-Detection Policy Tracker
Turnitin is a trademark of its owner. Genutext is independent and not affiliated with, endorsed by, or connected to Turnitin. Product behaviour described here reflects Turnitin's public documentation as of 28 September 2026; check the linked pages for current details.