Google Classroom does not detect AI writing. Its originality reports compare a turn-in with webpages, books and earlier school submissions, so they catch copied text, not AI text. To check for AI, paste the student's text into a separate detector. TextSight has no Classroom integration; below are the copy-and-scan route and the tools that plug in.
A lot of detector tools list LMS integrations they have not shipped. This page is the opposite of that. Here is exactly where a TextSight Google Classroom integration stands today.
Not built. No code has been written against the Classroom API, there is no TextSight add-on in the Google Workspace Marketplace, and we have not started Google's OAuth app verification, which an app reading student coursework would need. An integration is an idea on our list, not a scheduled release.
Three things, in order. First, enough waitlist replies to put Classroom ahead of Canvas, Blackboard, Schoology, Moodle and D2L Brightspace, which are also requested. Second, Google's OAuth verification for access to student coursework. Third, FERPA and DPA review with at least one district pilot before general availability. We are not publishing a date until the first of those is met.
Paste-and-scan on the web detector, the Chrome extension's side panel (paste the text in, or right-click selected text on pages that allow it), and Bulk Scan on Pro and above, which checks up to 50 texts in one job. It is not one-click, but it works today, and an integration would use the same detector.
Nothing below needs the planned integration. It is the copy-and-scan path teachers can use on a Classroom turn-in right now, with a real result from our live detector.
You are grading an essay that a student wrote in Google Docs and turned in through Classroom. One paragraph reads differently from the rest of the essay and from the student's earlier work. You want a second signal before you decide whether to raise it.
Open the student's attached Doc from the Classroom grading view. Select the paragraph you are unsure about, or the whole document, and copy it. If you prefer working from a file, use File, then Download, then Microsoft Word (.docx) in Google Docs to save a copy you can upload instead. The detector needs at least 50 words to return a score, so copy the full paragraph rather than a single sentence.
Paste the text into the free AI detector (no account needed for up to 3 checks a day and 1,000 words a scan), or sign in at app.textsight.ai for your plan's limits. You can also paste it into the TextSight Chrome extension's side panel so you stay on the grading tab; the extension needs a free account. The detector is trained on English and is not reliable on other languages. The workflow is the same one described on our AI detector for Google Docs page, because a Classroom essay is usually just a Google Doc with a due date.
Here is what a clear case looks like. On 31 October 2026 we ran a 113-word student essay paragraph, generated by an AI model, through the live detector. It scored 95% AI. A 102-word research abstract generated the same way also scored 95% AI. Note that 95 is the highest figure the headline shows, because the display caps there. The free plan stops at that overall score. On Starter and above, sentence-level highlights show which lines drove the result, which is the part worth bringing to a conversation.
A high score is evidence, not proof. In our published false-positive benchmark, the same detector read 1,180 human-written academic abstracts from 2018, years before ChatGPT. It labelled 74% of them human, left 20% uncertain, and wrongly flagged 69 as AI, a 5.8% false-positive rate. Across a full class, a few honest students can land in the flagged band.
Google Docs keeps a version history (File, then Version history). An essay that grew over several sessions looks different from one that appeared in a single paste. Put the score, the highlighted sentences, and the version history side by side, then ask the student to walk you through how they wrote it. If you leave a note in Classroom, a private comment keeps the question between you and the student. The decision stays with you and your school's integrity process.
These tools answer different questions. Some score the text a student submitted; others replay how the document was written. Knowing which is which matters more than picking a single winner.
| Tool | What it looks at | How it fits Classroom |
|---|---|---|
| Google Classroom originality reports | Matching text in webpages, books and (if an admin turns on school matches) earlier school submissions. No AI-writing check is described. | Built in. Fundamentals: 5 assignments per class; unlimited with Teaching and Learning or Education Plus. Students can run 3 reports before turning in. Google help, checked 2 Nov 2026. |
| TextSight | The submitted text. Free returns one overall AI score; sentence-level highlights from Starter. | No Classroom integration. Copy and paste into the web detector or the Chrome extension side panel, or upload a downloaded file. |
| Pangram | The submitted text, as an AI-detection report. | A Google Classroom add-on in the Workspace Marketplace: you attach an AI usage policy to an assignment and open the report while grading. Marketplace listing, checked 2 Nov 2026. |
| Originality.ai | AI detection, plagiarism and a writing replay. | A Chrome extension whose panel appears in the Classroom grading view, not a Classroom add-on. Originality.ai announcement, checked 2 Nov 2026. |
| Turnitin | Submitted work, for schools that license it. | Licensed to institutions and not sold to individuals, so access depends on what your school has set up. See our Turnitin alternative notes. |
| GPTZero | The submitted text, as a classifier, plus a writing replay. | Lists a Google Classroom integration and a "GPTZero for Education" Classroom app in the Workspace Marketplace. Its pricing page does not say which plan includes it. GPTZero integration page and Marketplace listing, checked 2 Nov 2026. Compared on GPTZero vs TextSight for teachers. |
| Brisk Teaching | How the document was composed: typing, revisions and paste events. Brisk says it "doesn't produce an accuracy score". | Inspect Writing runs inside Google Docs through the Brisk Chrome extension (Brisk, checked 2 Nov 2026). More on our Brisk comparison. |
| Draftback | A replay of a Google Doc's revision history. It does not score text. | A Chrome extension for Google Docs. |
Process replay and text scoring work well together. Replay needs a live Doc with history; a text score works on anything you can copy, including pasted turn-ins and uploaded files. Classroom's own originality reports look for matching sources, which is a plagiarism check, not an AI-authorship check. For a wider ranking of detectors for schools, see best AI detector for teachers and best AI detector for edtech.
Nothing below is built. The scope is a working idea, not a feature promise or a date. Items marked Planned are what a first version would aim for; items marked Out of scope are deliberately excluded.
| Capability | v1 status | Notes |
|---|---|---|
| Pull student turn-ins via Classroom API | Planned | Read-only access to student coursework, teacher-authenticated only. |
| Run AI detector on each submission | Planned | Same detector and model as app.textsight.ai. No separate calibration. English-trained. |
| Run Plagiarism Risk check | Planned | Bundled in the same pass, no extra teacher click. |
| Write result as private grader-side comment | Planned | Visible only to the teacher, not the student. Includes a TextSight report link. |
| One-click bulk scan for a whole assignment | Planned | Process all submissions for an assignment with one teacher action. |
| Per-assignment scan history in TextSight | Planned | Teacher-side log of what was scanned, scored, and when. |
| CSV export of class-level results | Planned | For teachers who keep a separate gradebook outside Classroom. |
| Sentence-level highlights in the report link | Already in detector | The web report already shows per-sentence colour-coded highlights on Starter and above. Free shows the overall score. |
| Auto-apply grades based on AI score | Out of scope | By design. Detector signals inform a teacher conversation, not a grade. |
| Notify students of their AI score | Out of scope | Teachers told us this should be a private signal, not a public flag. |
| Auto-flag or auto-reject submissions | Out of scope | No automated discipline events. Ever. |
| Student-side appeal portal | Out of scope | Belongs with the school's existing academic-integrity process. |
| SSO with Google Workspace for Education | Planned via Google OAuth | Teacher signs in with the same Google account they use for Classroom. |
| Admin-level deployment across a district | Post-v1 | v1 is teacher-installable. District-wide deployment ships after pilot feedback. |
| Real-time scanning during typing | Out of scope | Scanning happens on turn-in, not during the student's writing session. |
This scope is set against TextSight's existing detector accuracy. Nothing about the integration changes the underlying model: same calibration, same per-sentence highlights, same false-positive rate.
These are the grading situations where copy-and-scan is slowest, and where a native integration would save the most time.
Teachers grading 90 to 150 Classroom turn-ins in one Friday afternoon do not have 20 extra seconds per submission to copy, switch tabs, paste, scan, switch back, and write a note. An integration's whole purpose would be collapsing that loop to a single grader-side action so the AI signal is available without breaking grading flow.
When one essay scores high on AI, the immediate next question a teacher asks is whether it is a class-wide pattern. A per-assignment view would show the score distribution across all submissions, so the teacher could tell at a glance whether to address one student or rebrief the whole class on AI use.
Classroom assignments often have co-teachers. A private comment on the submission would let a co-teacher see the same AI signal without a second scan, which is hard to do cleanly with paste-and-scan.
Many teachers want a private nudge to bring up AI use in a one-on-one conversation, not an automated flag that triggers an integrity process. A grader-side comment would suit that conversation. No auto-grading, no auto-notification.
Department heads and curriculum directors want quarter-over-quarter AI-usage trend data without seeing any individual student. An aggregated, anonymisable CSV export would be the way to meet that ask. It is part of the idea, not a built feature.
FERPA is the school's compliance posture, not a vendor checkbox. TextSight's job is to be a clean upstream partner so the school can defend the deployment. Here is how TextSight handles text today, and how a Classroom integration would work.
Know this before you paste student work: when a teacher is signed in, each scan and its text are saved to that teacher's scan history by default, so the report can be reopened. Free accounts see the last 7 days of history; paid plans see all of it. Switching off “Store my scan history” in Settings stops new scans being saved, and saved scans can be deleted. The details are in our content-handling notes.
School districts and higher-education institutions can request a Data Processing Addendum. Email support@textsight.ai with "DPA request" and the institution name to get the current draft, and see our trust page for how procurement teams request one.
The idea is for an integration to run under the teacher's own Google sign-in, read-only, with no separate student data export, no shadow accounts and no student sign-in. None of this is built.
TextSight's current subprocessor list is published on our trust page. A Classroom integration would add calls to Google's own Classroom API on the teacher's session.
Run the integration through your existing third-party application review (Common App Review for K-12 districts, or your institution's IT security review for higher ed) before deploying it more broadly. We will provide the architecture diagram, the DPA, and the subprocessor list. The "is this FERPA okay" answer always belongs to the school, not the vendor.
There is no signup form behind this. Email us with the details below and we will count your request when we decide which LMS to build for first.
Email support@textsight.ai with the subject line "Google Classroom waitlist" and please include:
We read every reply and use them to rank Google Classroom against Canvas, Blackboard, Schoology, Moodle and D2L Brightspace. No marketing email list, no auto-drip. We will write back if the build starts.
Prefer to talk first? Ask for a short call in your email.
Every LMS integration on this list is in the same planning state as Google Classroom. The waitlist signal across all six is how we pick which one ships first.
Planned SpeedGrader integration. Read-only courseworks scope. Same v1 scope as Classroom.
See the roadmap ›Planned Grade Center integration via the Learn REST API. Higher-ed focused v1.
See the roadmap ›Planned local plugin for self-hosted Moodle instances. Open-source-friendly install.
See the roadmap ›The educator hub. Detection methodology, ESL fairness notes, and how teachers use TextSight today.
Open the hub ›K-12 and higher-ed workflow notes. How to read scores, what to do on a flag, how to talk to a student.
Read the guide ›The other big classroom integration request. Same planning-phase status, same waitlist process.
See the roadmap ›The higher-ed integration request. REST API, Grade Center, same planning-phase status.
See the roadmap ›The underlying detector that the integration will use. Calibration, ESL fairness notes, benchmark dataset.
Read the methodology ›Free, Starter $9.99, Pro $19.99 ($14.99 annual), Business $39.99 ($29.99 annual).
See pricing ›Tell us how your classes use Classroom. Until anything is built, the web detector and Chrome extension cover the paste-and-scan path.