Home / Guides

A free AI detector for teachers, and how to use it fairly

If you teach, you've probably read an essay recently and thought, "Did a student write this?" A detector can help you decide whether to look closer. It can't tell you what happened. Here's how to use one without hurting a student who did nothing wrong.

Updated October 11, 2026 · By the IsThisAI? team

Checking a submission

With IsThisAI?, you paste the text or upload a .txt, .docx, or .pdf file and get a score with the most AI-like sentences highlighted. The check runs in your browser, so the student's text isn't sent to us or stored. That matters: Vanderbilt warns that many third-party detection tools have unknown privacy and data-use policies, so entering student work into them may conflict with student privacy protections such as FERPA (Vanderbilt).

Before you run anything, a few habits make the result more useful:

• Check enough text. A few paragraphs of prose at minimum. Turnitin won't score anything under 300 words of prose, because results on short text are unreliable (Turnitin FAQ).
• Skip lists, outlines, and resumes. Turnitin ignores bullet points entirely (Turnitin FAQ), and in our own testing, none of six free detectors could reliably catch AI-written resumes.
• Know the student's baseline. An in-class writing sample from early in the term is the best thing to compare against. A sudden change in voice tells you more than any score.

Reading the result

A detector score is a probability from a statistical model, not a measurement. Some practical ways to read it:

• Middle scores mean "unclear." Anything in roughly the 30% to 70% range is the model shrugging.
• Look at what's highlighted. Bland intros and summary conclusions get flagged a lot, because lots of people write them the same way. Highlights in the body, where the student's actual thinking should be, are more meaningful.
• Expect disagreement between tools. Different detectors are different models trained on different data. A 14-tool study found the tools they tested were neither accurate nor reliable, especially on edited or paraphrased text (Weber-Wulff et al., 2023).
• Think about who wrote it. Detectors flag non-native English writers far more often. In a Stanford study, seven detectors labeled more than half of 91 human-written TOEFL essays as AI, with an average false positive rate of 61.3% (Liang et al., 2023).

Why a score isn't proof

Even the biggest vendor says this. Turnitin's documentation states its model "may not always be accurate" and "should not be used as the sole basis for adverse actions against a student" (Turnitin Guides). Its own target is a document-level false positive rate under 1% for documents with over 20% AI writing, which still means roughly one human-written paper in a hundred could be flagged (Turnitin FAQ).

Universities have taken that seriously. Vanderbilt disabled Turnitin's AI detection in 2023 and noted that, at its volume of 75,000 papers a year, a 1% false positive rate could mean around 750 wrongly flagged papers (Vanderbilt). The University of Pittsburgh's Teaching Center recommends against using AI detection tools, because they aren't accurate enough to prove a student violated an academic integrity policy (Pitt Teaching Center). The University of Iowa's teaching office goes further and says instructors "should refrain from using AI detectors on student work" (University of Iowa). MIT Sloan's teaching team gives similar advice (MIT Sloan).

So use a detector the way you'd use a hunch: as a reason to look closer and ask questions, not as a finding.

Talking to the student

The conversation is where you actually learn something. A few approaches that tend to go well:

1. Lead with curiosity, not accusation. "Walk me through how you wrote this" opens a very different conversation than "A detector says you used AI."
2. Ask about the content. Why did they choose that source? What did they mean in paragraph three? What would they change? People who wrote a piece can usually talk about it.
3. Ask for process evidence. Google Docs or Word version history, outlines, notes, and earlier drafts. Our guide on how students can prove they wrote something lists what to look for.
4. Share the report. If you used Turnitin, students can't see the AI report themselves, but you can download it and share it (Turnitin FAQ). Letting them see exactly what was flagged is fair.
5. Follow your institution's process. If you still have concerns, use the formal academic integrity process rather than deciding alone.

A classroom policy template

Most confusion comes from unclear rules. A short, specific policy at the top of the syllabus prevents a lot of it. Here's a starting point you can adapt:

Pair the policy with assignments that make the process visible: in-class drafting, short reflections on how a paper developed, or a quick conversation about a draft. Iowa's teaching office recommends exactly this: clear expectations, a focus on the writing process, and open conversations about acceptable AI use, instead of detection (University of Iowa).

For background on the numbers, see are AI detectors accurate? and AI detector false positives.

More guides

Sources

1. Turnitin, "Turnitin's AI writing detection capabilities FAQs" (Turnitin Guides, updated Oct 7, 2026). https://www.turnitin.com/products/features/ai-writing-detection/faq
2. Turnitin, "Using the AI Writing Report" (Turnitin Guides). https://guides.turnitin.com/hc/en-us/articles/22774058814093-Using-the-AI-Writing-Report
3. Liang, Yuksekgonul, Mao, Wu & Zou, "GPT detectors are biased against non-native English writers," Patterns 4(7), 2023. https://www.cell.com/patterns/fulltext/S2666-3899(23)00130-7
4. Vanderbilt University Brightspace team, "Guidance on AI Detection and Why We're Disabling Turnitin's AI Detector" (Aug 16, 2023). https://www.vanderbilt.edu/brightspace/2023/08/16/guidance-on-ai-detection-and-why-were-disabling-turnitins-ai-detector/
5. Vanderbilt University, "Academic Integrity and Generative AI". https://www.vanderbilt.edu/generative-ai/academic-integrity/
6. University of Pittsburgh Teaching Center, "Encouraging Academic Integrity" (statement about AI detectors). https://teaching.pitt.edu/resources/encouraging-academic-integrity/
7. University of Iowa Office of Teaching, Learning, and Technology, "The case against AI detectors" (Sept 30, 2024). https://teach.its.uiowa.edu/news/2024/09/case-against-ai-detectors
8. MIT Sloan Teaching & Learning Technologies, "AI Detectors Don't Work. Here's What to Do Instead.". https://mitsloanedtech.mit.edu/ai/teach/ai-detectors-dont-work/
9. Weber-Wulff et al., "Testing of detection tools for AI-generated text," International Journal for Educational Integrity 19, 26 (2023). https://link.springer.com/article/10.1007/s40979-023-00146-z

This guide is general information, not legal or academic-policy advice. AI detection results, including ours, can be wrong.