For Teachers, Professors & Institutions

AI Detection for Education

Check student writing for AI-generated content — with sentence-level analysis, PDF reports for your records, and privacy options built for academic settings.

Sentence-level AI scoresPDF reports per submissionZero-retention optionWorks alongside any LMS
1 free detection remaining

How It Fits Your Grading Workflow

No installation required — it works alongside Canvas, Moodle, Blackboard or any LMS you already use.

📥

1. Students Submit

Students hand in essays through your LMS as usual — nothing changes for them

📄

2. Paste or Upload

Paste the text or upload the document (PDF/DOCX) — most instructors spot-check borderline work

🔬

3. Review Evidence

See which sentences score as AI-written, with a confidence score per sentence — not just one number

🗂️

4. Keep Records

Generate a PDF report per submission for academic-integrity documentation

Looking for a native Canvas (LTI 1.3) integration? We're scoping this with institutional design partners — talk to us.

Built for Student Privacy

Student writing is sensitive. Our defaults are minimal — and institutions can go further.

🔍 Minimal by Default

Submitted text is processed for analysis. Your account history keeps only the result and a short snippet — never the full document.

🛡️ Zero-Retention Option

For institutional agreements we offer zero-retention processing: no snippets and no per-submission results are stored — analysis happens in-flight, results go only to you.

📁 You Control Exports

PDF reports are only created when you explicitly generate them — they're your documentation, kept until you delete them.

⚖️ A Note on Responsible Use

No AI detector — ours included — is definitive proof of AI authorship. False positives happen, especially with very formal writing and non-native English speakers.

That's why we show sentence-level evidence instead of a single verdict: detection results should start a conversation with the student, not end one. We recommend using them as one indicator alongside your own judgment and your institution's academic-integrity process.

Institutional Pilots & Partnerships

Writing programs, departments and integrity offices: we offer evaluation accounts for instructors, volume pricing, zero-retention agreements — and we're selecting design partners for our LMS (Canvas LTI 1.3) integration.

Direct line to the founder — replies within one business day

Educator FAQ

Do you integrate with Canvas, Moodle or Blackboard?

There is no native LTI integration yet — instructors use the web dashboard alongside their LMS (paste or upload submissions). A Canvas LTI 1.3 integration is on our roadmap and we are actively looking for institutional design partners to build it with. Our REST API is available today for custom integrations by your IT team.

How should I handle a flagged essay?

Treat the result as evidence for a conversation, not a verdict. Review the sentence-level scores, compare with the student's previous writing, and follow your institution's academic-integrity process. False positives occur — particularly with formal academic prose and non-native speakers.

What happens to the student text I submit?

By default, text is processed for analysis and only the result plus a short snippet is kept in your account history. Institutions can enable our zero-retention option, where no snippets and no per-submission results are stored at all.

Is there educator or institutional pricing?

Yes. Individual instructors can start free and upgrade to unlimited checking for $9.99/month. For departments and institutions we offer volume pricing, evaluation accounts and formal agreements — contact us for a pilot.

Can I check many essays at once?

You can upload documents (PDF/DOCX) one at a time today, each with an optional PDF report. Bulk upload for whole course sections is in development — pilot partners get early access.

Which AI models can you detect?

Our detector is trained on output from ChatGPT, GPT-4-class models, Claude, Gemini and other current writing assistants, and is continuously updated as new models appear. Longer samples (300+ words) give the most reliable results.