Inquiries Regarding Turnitin AI Detection Reports
Hello Everyone,
I am currently reviewing students’ files and uploading them to Turnitin to detect both plagiarism and AI-generated content.
According to International Baccalaureate (IB) policies students are permitted to use such tools provided that they are properly acknowledged and cited according to a recognized citation style such as MLA or APA.
However, I am facing several issues with the Turnitin AI report and I would appreciate your guidance regarding the following points:
1. In some cases I found that the report still identifies certain sections as AI-generated even though the student has already properly cited and acknowledged the use of AI for that specific part, What is the appropriate solution in this case?
Also unlike the similarity report there does not seem to be a way to exclude these cited sections from the overall AI percentage report.
2. Another issue concerns students in Business Management where they may evaluate or analyze articles that themselves were written using AI writing tools and they properly cite these sources Despite this those parts still appear as a high AI percentage in the report and again I do not seem to have the ability to exclude them.
3. Some reports display an asterisk (*) which may indicate that the AI score is below 20% but it may also suggest that some students are using AI-to-human rewriting/paraphrasing tools to mask AI-generated writing.
At the moment I am unable to clearly distinguish between these two cases.
How can I identify this accurately?
4. Finally, I would like to ask about Arabic language support.
So far Arabic has not yet been included among the languages that the program can reliably assess for AI writing detection.
Will this feature be available soon or is there any expected timeline for its release?
Thanks in advance! I look forward to your response.
2 replies
-
Hi Tamer, nice to hear from you again!
Sometimes, sections may be flagged as AI-generated. This can happen in text that is highly polished, uniform in structure, or contains repeated phrases. It’s always best to review results in context and consider your knowledge of the student’s work.
Gailene from our previous #AskTurnitin series shared some insight into how AI detection works:
“While we are continuously monitoring our tool's efficacy to keep our False Positive Rate (FPR) below 1%, the tool can make mistakes. Some AI content generators overlap more closely with human writing, which can occasionally lead to incorrect flags. Text that is less complex, very uniform, or perfectly grammatical can also trigger potential AI signals. Even cited content may be flagged if sentences are short or highly structured. All predictions come with a spectrum of confidence, and we’re actively researching ways to make the detection report easier to interpret. We’re excited to deliver more usability updates in the coming months!”
For practical guidance on reviewing AI detection results in context, we recommend checking our instructional resource: “Diving Deeper into AI Writing at Turnitin – for Instructors” as well as "AI is here to stay in the classroom, so why do we need AI detectors?"
Now, regarding Arabic language support, it is not yet available for AI writing detection, but we’re actively working on expanding language support. Arabic will be coming later this year, so stay tuned!
- Kat, Turnitin Team