How can teachers tell if a student used AI?

The most reliable way to tell whether a student used AI is to compare the work with what you already know about that student, then ask them about it. Detector scores alone are not proof. Teachers who know a child's handwriting, spelling habits and way of explaining things are usually better placed than any tool.

Signs exam boards tell teachers to look for

The UK's Joint Council for Qualifications publishes guidance for teachers, AI Use in Assessments, with a list of indicators. The ones that matter most in primary and middle school:

• A different voice. The style, vocabulary or sentence length does not match the student's classwork, or it shifts partway through.
• Generic content. The answer is fluent but not tied to the specific task, the class discussion or local details.
• Sources that do not exist. References that cannot be found, or invented quotes attributed to real people.
• Confident errors. Mistakes or non sequiturs stated with total certainty inside otherwise polished text.
• Leftover AI phrasing. Disclaimers such as "As an AI" or "I cannot browse the internet".
• Unexpected spellings. American spellings from a UK student (or the reverse) when they normally write otherwise.
• Typed work from a student who normally handwrites, or repeated concluding statements and template-like structure.

None of these proves AI use on its own. A child might have had help from a sibling, copied from a website or simply had a good day. Treat them as reasons to look closer.

Why AI detectors are not enough

• OpenAI withdrew its own AI text classifier in July 2023, saying it was no longer available "due to its low rate of accuracy".
• A Stanford study published in Patterns found that popular GPT detectors flagged essays by non-native English writers as AI-generated far more often than essays by native speakers, and the authors cautioned against using detectors in educational settings.
• The JCQ guidance itself says detection tools can support a teacher's judgement but should be part of a holistic review, not relied on alone.

For a deeper look at the tools, see Does AI detection in schools actually work?.

What to do when you suspect AI use

1. Gather your own evidence. Put the piece next to two or three recent pieces of in-class work.
2. Have a curious, private conversation. Ask the student to talk you through their answer: "What was the hardest part to write? Where did you find this fact? What would you change?" A child who wrote the work can usually explain it.
3. Ask for a short redo in class. A paragraph on the same topic, written without devices, tells you a lot.
4. Separate learning from discipline. For younger children especially, AI use is often confusion about the rules rather than deliberate cheating. Clarify the expectation first.
5. Follow your school's policy and record what you did, especially for assessed work.

Make it easier to see the learning

The more of the process you can see, the less you have to guess. Short drafts in class, planning sheets and oral explanations make AI-written work stand out and help honest students show their thinking. How do I stop students cheating with AI? covers task design in more detail.