CMDLearn

Learn

Do AI detectors work on homework? What parents and teachers in India should know

CMDLearn ·

In short: no AI detector can prove that a child's homework was written by ChatGPT. The companies that make these tools say so themselves: OpenAI withdrew its own detector in 2023 because it was not accurate enough, and Turnitin, a plagiarism checker that also checks for AI writing, says its score must not be the only reason for action against a student. Detectors also flag real work by mistake, and one study found this happens far more often to people writing in English as a second language. A score is a reason to talk to the child, not proof. This page explains what the makers admit, and what to do if a child's work is flagged.

What the makers of detectors say about their own tools

OpenAI released a tool in January 2023 to say whether a text was written by a person or by AI. Its own announcement called it "not fully reliable": on OpenAI's test set it caught 26% of AI-written texts, and it labelled human writing as AI-written 9% of the time. OpenAI said it was very unreliable on texts under 1,000 characters and recommended it for English text only. On 20 July 2023 OpenAI took it down, giving "its low rate of accuracy" as the reason.

Turnitin, which offers an AI-writing check alongside its plagiarism checker, states these limits in its own guides:

Many free "AI homework detector" websites publish no such limits at all. If a tool does not tell you how often it is wrong, you cannot know what its score means.

How often real work gets flagged

The numbers above are each company's own. Independent research adds a warning that matters in India. In a 2023 study in the journal Patterns, researchers at Stanford University ran seven popular AI detectors on 91 essays written for an English-language exam (TOEFL) by non-native English writers, and on 88 essays by US eighth-grade students. The detectors judged the US students' essays correctly, but labelled more than half of the non-native writers' essays as AI-generated: the average false-positive rate was 61.3%. Every one of the seven detectors flagged 19.8% of these human-written essays, and at least one detector flagged 97.8% of them.

The researchers' explanation: many detectors treat plain, predictable wording as a sign of AI, and people writing in a second language often use simpler words and sentences. They "strongly caution against the use of GPT detectors in evaluative or educational settings", especially for non-native English speakers.

The study did not test Indian students, and detectors have changed since 2023. But many children in India write their English homework in a second or third language, often in short, careful sentences. That is exactly the kind of writing the study found most likely to be flagged wrongly.

Do CBSE or Indian boards use AI detectors?

We found no CBSE circular on AI detectors or on checking homework for AI (we read the titles of CBSE's 2026 academic circulars on cbseacademic.nic.in on 7 November 2026; the AI ones are about the Computational Thinking and AI curriculum and an AI literacy course). So whether a detector is used, and what happens after a flag, is up to each school. Ask for the school's policy in writing if your child's work is questioned.

If a teacher says your child's homework was written by AI

Stay calm, and ask for the evidence in full: which tool, what score, and for which part of the work. Then:

  1. Show how the work was made. Rough work in a notebook, earlier drafts, notes, the book pages your child used, or, if it was typed in an online document, its version history (the record of every change and when it was made).
  2. Let your child explain it. Ask the teacher to talk to your child about the work: what it means, why a word or step was chosen, what they would change. A child who wrote it can usually explain it.
  3. Mention the tool's own limits. Turnitin's guidance says its score should not be the only basis for action against a student, and its AI check does not cover Hindi, Marathi or other Indian languages. A short answer under 300 words is below its minimum.
  4. Ask what the school's rule is. If the school allows some uses of AI (for example, to explain a topic), check whether your child stayed within them.

If your child did use a chatbot to write the work, treat it as a learning problem, not only a discipline one. Our guide on how to use ChatGPT for homework without cheating shows honest ways to study with it.

For teachers: checking work without a detector score

For tools that help with the rest of your workload, see AI tools for teachers in India, and for classroom routines, our practical guide to AI in the Indian classroom.

Where CMDLearn fits

CMDLearn is building an AI tutor for Classes 1 to 10 in Hindi, Marathi and English: it explains one step, checks the child's work and leaves the writing to the child. It is still being built and no school or child uses it yet. If you would like to hear when it opens, join the waitlist.

Sources

All read on 7 November 2026. openai.com and guides.turnitin.com refuse our fetch, so their pages were read in Internet Archive copies, as noted.

Tool names are given as examples; we are not paid by any of them. This is general information for parents and teachers, not legal advice; follow your school's policy.