Shipping the engine, layer by layer.
Noplag's detection engine is open source, built in public. Today it finds copied and lightly-edited text across billions of sources, plus a real-time web check. Paraphrase, meaning-based, and AI-writing detection are on the roadmap below — each shown honestly by status.
What the engine catches
The detection layers. Verbatim and near-verbatim ship today; paraphrase, meaning-based, and AI-writing detection are on the way.
Exact-match detection
Finds text copied verbatim from any source.
Near-match detection
Catches lightly-edited copying — reordered words and small swaps.
Paraphrase detection
Detects reworded passages that keep the original sentence structure.
Meaning-based detection
Catches full rewrites and translated paraphrase by meaning, not wording.
AI-writing detection
Flags likely AI-generated text — no source needed.
Where we search
The corpora every check runs against.
Open web corpus
Billions of web pages (Common Crawl / FineWeb).
Wikipedia
The full encyclopedia corpus.
Academic papers
Scholarly corpus (S2ORC).
Live web search
A real-time web check at submission (Premium).
Noplag Database
Anonymized fingerprints of documents checked before — across all tiers, opt-out.
Your private library
Check against your own uploaded documents (My Folders).
Expanded sources
Legal, patents, books, and biomedical corpora.
Where you can use it
Web app
Full check, report, and sharing in the browser.
Mobile apps (iOS / Android)
Scan and check on the go.
Developer API
Programmatic access on paid plans.
Desktop app
A native desktop client.
Browser extension
Check from any page.
The engine itself
Open-source engine
The detection engine, going public on GitHub — read it, fork it, self-host it.
Watch the engine come together.
Every layer ships in the open. Star the repo to follow along — paraphrase, meaning-based, and AI-writing detection land here as they're built.
Status and incidents live on status.noplag.com — separate from this roadmap.