Plagiarism checker for educators.
Bulk uploads for full classes. Per-student dashboards. Canvas, Moodle, Blackboard, and Brightspace integrations. Reports that respect proper citations, plus an AI detector — coming soon — that's honest about its limits, including for your ESL writers.
Four educator workflows. One account.
Spot-check a quiz or essay in 30 seconds
Free tier covers a full classroom. No LMS required.
- Paste a student response, see overlap immediately
- Mark properly cited passages as expected
- Print or share a clean PDF report
Bulk-grade assignment submissions
Run 100+ submissions through one queue.
- Drop a CSV or zip of .docx / .pdf submissions
- Per-student dashboard with similarity + AI score
- Click into any flagged interval to read the source
Set a citation policy, watch adoption
Org-level dashboard for academic integrity.
- Roster sync from Canvas / Moodle / Blackboard / Brightspace
- Aggregate similarity trends, no per-student profiling
- Export audit logs for accreditation review
Investigate a contested score with the receipts
Read the algorithm, then read the report.
- X-Engine-Commit on every report — reproducible
- Override AI verdict with documented reasoning
- Apache 2.0 engine — defensible in a hearing
One queue. The whole class through it.
Drop a CSV or a zip of student submissions. Get per-student reports back. No per-document upload, no per-submission billing surprises.
Roster-load or drop submissions
Pull a class roster from your LMS, or drop a CSV / zip of .docx / .pdf submissions. Up to 250 documents per batch on Premium, unlimited on Enterprise.
One queue, parallel checks
The full class runs in parallel. Each submission goes through the same retrieval cascade (winnowing → live web today; MinHash and vectors as they ship). Median 42s per document.
Per-student dashboard
Sortable similarity + AI score columns. Click any row for the full highlighted report, the source URLs, and the engine commit that produced the score.
Four LMSes. One install. Single-sign-on per student.
LTI 1.3 across the four largest higher-ed and K-12 LMSes. Install once at the school level; teachers add Noplag as an external tool to any assignment.
External tool in Assignments. Roster + grade passback supported.
- Roster sync at section level
- Originality + AI score writeback to gradebook
- Speedgrader-compatible inline annotations
Plugin from the Moodle Plugins Directory. SSO via OAuth2.
- Cohort + group roster sync
- Score writeback to gradebook
- Self-host compatible (works with on-prem Moodle)
LTI 1.3 + Building Block. Sandbox-tested with Anthology releases.
- SafeAssign-compatible upload format
- Roster sync from courses
- Inline report in Bb Annotate
Approved external tool. Provisioned at org-unit level.
- Brightspace Lessons embed
- Turnitin migration importer
- K-12 + higher-ed deployments
One class. One screen. Sortable everything.
Per-student row with similarity, AI score, and a click-through to the highlighted report. No PDF-export click hunt — that's a keyboard shortcut.
A signal, not a verdict. Especially for your ESL writers.
AI detectors over-flag non-native English writers
Liang et al. 2023 measured a 61.3% false-positive rate against ESL writers on detectors that scored 5.1% on native writers. The bias source is well-understood: ESL writing patterns share statistical signatures (lower perplexity, more formulaic phrasing) with AI-generated text. We rebuilt the AI verdict to acknowledge this.
We calibrate, then we annotate — and we never auto-fail
When AI detection rolls out, submissions detected as likely ESL (lexical signature + optional roster-flagged status) will have the AI score recalibrated against the ESL-only validation set. The verdict label will shift to ESL ADJ instead of LIKELY when the recalibration disagrees with the native-detector threshold. Residual error is ~3% — not zero. The score is reported, the override path is documented.
A meeting, not a default-fail
AI detection will ship with a per-submission discussion guide for educators: the score, the methodology, the false-positive rate for the student's profile, and questions to ask before escalating to an integrity panel. The score is one input. The student's process — drafts, sources, citation hygiene — is another. The verdict is yours, not ours.
A passage in quotation marks with the citation under it isn't plagiarism.
Properly attributed quotes, paraphrases with in-line citations, and standard bibliographic conventions get marked as expected — not flagged red. We classify each matched interval (Direct quote / Paraphrase / Common knowledge / Unattributed) before computing the headline similarity number. The teacher view shows the classification next to every interval, so a 28% similarity score on a heavily-cited literature review reads differently from a 28% score on an uncited paraphrase. The teacher's judgement still rules — but the report doesn't punish a student for following MLA.
Read the rubric notesThe questions a department chair actually asks.
- What does “free for K-12 teachers” actually cover?
- Unlimited class checks on the Free K-12 tier for any teacher with a verified .edu / school-domain email. Up to 35 students per class, up to 25 active classes per teacher. No card. No usage cliff. Higher-ed lecturers start at the Premium plan ($79/mo) which adds bulk-batch + LMS integrations.
- Do you store student submissions in a shared database?
- By default, submissions are added to the Noplag Database (the shared cross-customer index) and never used to train anything. You can opt out per institution to keep submissions in the teacher's own folder only, and the shared index is bypassed entirely on K-12 deployments. Right-to-erasure on every fingerprint we hold.
- What if my ESL student gets flagged as AI?
- AI detection is coming soon (on the v1.2 roadmap). When it ships, the AI score will be reported with an ESL-adjusted band when the lexical signature suggests non-native English (or the roster flags the student as ESL), and the recalibrated verdict will carry a documented residual error rate. The teacher view will ship with a discussion guide for the conversation — the score is one input, the student's drafting process is another. Never an auto-fail.
- How do reports treat properly cited quotes?
- Every matched interval is classified before it touches the headline similarity score: Direct quote (with attribution), Paraphrase (with / without citation), Common knowledge, Block quote, Bibliography. Properly cited intervals are excluded or down-weighted. The classification is shown next to the interval — a 28% similarity report on a literature review reads differently from a 28% score on uncited paraphrasing.
- Which LMSes do you integrate with and how?
- Canvas, Moodle, Blackboard Ultra, Brightspace via LTI 1.3. Install once at the institution level; teachers attach Noplag to any assignment from the LMS picker. Roster sync, gradebook writeback, inline annotations in Speedgrader / Bb Annotate / D2L. A 15-minute call with school IT gets the install live.
- Can I import my Turnitin assignment library?
- Yes — the Brightspace and Canvas connectors ship with a Turnitin migration importer that maps assignment definitions + rubrics. Historical submissions don't transfer (Turnitin doesn't expose them), but new submissions on the same assignment ID flow into Noplag from the migration date forward.
- What's the data residency story for EU schools?
- EU residency endpoint (api.eu.noplag.com) provisioned on Pro and above. Submissions never leave EU infrastructure. GDPR-compliant by default; DPA available on request. Self-host the engine entirely on your own servers if your district policy requires it — the Apache 2.0 reference is identical to the cloud.
- How do you handle academic-integrity hearings? Is the report defensible?
- Every report stamps X-Engine-Commit and a corpus snapshot timestamp — the same check is reproducible months later. The engine is Apache 2.0, so the algorithm is reviewable in an integrity panel. Override logs (who marked an interval as expected, when, why) are exportable. Designed to survive a hearing, not just to produce a number.
- What languages do you support for ESL classrooms?
- Detection is calibrated for English, Spanish, Portuguese, Polish, and Ukrainian at launch, with more languages rolling out. Detection quality is benchmarked on PAN-PC-11 and published at /docs/developers/benchmarks; per-language evaluation is on the roadmap.
- What if a student disputes the AI score?
- AI detection is coming soon (v1.2 roadmap). Once live: teacher view → click the student row → “Override verdict” with a freeform reason. The override is logged with the engine commit and corpus snapshot of the original check. The discussion guide that ships with every flagged submission is designed for exactly this conversation. We never auto-fail; the verdict is yours.
Free for K-12. Set up your class in five minutes.
No card on signup. K-12 teachers get unlimited free class checks. Higher-ed pilots start at $79/mo per teacher. School-wide LMS install needs a 15-min call with IT — we'll help you stand it up.