Skip to content
Global 100 Forum

Does Grammarly have an AI detector, and how accurate is it?

Asked by Colm B. · Detector tools compared2 repliesupdated

Short answer

I run the information literacy sessions in a secondary school library in Ireland. Grammarly is installed on a lot of the pupils' laptops, and last week a Transition Year pupil showed me a result from Grammarly's AI detector on her history project: a percentage of the text that "appears to be AI-generated". She wrote it herself, mostly in the library where I could see her, and she was upset.

Written by Colm B. ·
CiteGlobal 100 Forum, "Does Grammarly have an AI detector, and how accurate is it?", https://global100.org/forum/q/does-grammarly-have-an-ai-detector-and-how-accurate-is-it/, accessed 2026-11-18.
  1. Member ·

    I did not know Grammarly even had a detector. Now I need to explain it to a class of fifteen-year-olds without scaring them or overselling it.

    So, three questions. Is it a separate tool or part of the normal Grammarly checker? What exactly is the percentage a percentage of? And how accurate is it, given the page says it ranks first on some benchmark? I like to give pupils a short checklist, so anything concrete helps.

    0
    ReplyLink

Plain text or simple Markdown. Links are nofollow. Your email is never shown.

  1. Aisha R.Member ·

    I have used it on my own essays before submitting, mostly out of nerves. The free version gives you one number and that is it. If you sign in with a paid plan you get more, like which phrases it thinks look like AI and a button to rewrite them. I found the rewrite button more stressful than helpful, to be honest, because it makes you feel your normal sentences are wrong.

  2. Hana SatoMember ·

    On the benchmark claim: RAID is a real academic benchmark, built by university researchers, and it is a good one. But a leaderboard rank tells you how a detector did on that dataset at a fixed setting. It does not tell you how often it will be wrong on a fifteen-year-old's history project, which is a very different kind of text from most benchmark data.

Global 100 Forum. Questions and answers on AI detection, deepfakes and content authenticity.