About Plagum
Clearer checks. Better context.
Plagum helps students and writers review plagiarism similarities and probabilistic AI signals without treating two different checks as one conclusion.
Our approach
A useful report should help you understand what it found.
Plagiarism checking and AI detection answer different questions. Plagiarism checking looks for similarity with available sources, while AI detection estimates patterns associated with AI-generated writing.
Plagum keeps those results separate and gives you the context needed to review them. The goal is not to turn an automated score into a final judgment, but to make the underlying report easier to understand and use responsibly.
How we work
Four principles behind Plagum.
Context before conclusions
A similarity percentage is more useful when you can inspect the passages and available sources behind it.
Separate checks, separate meanings
Plagiarism checking looks for source similarity. AI detection estimates AI-associated writing patterns. Plagum does not merge them into one score.
Clear limits
AI indicators are probabilistic and can be wrong. We explain that directly instead of presenting an automated result as proof of authorship.
Your text stays under your control
Submitted documents are not used to train AI models and are not automatically added to the general comparison database.
Plagum in practice
The product, without inflated numbers.
These are practical limits and capabilities you can actually use.
- 70+
- languages supported for plagiarism and AI checks
- 24 MB
- maximum supported size for each uploaded file
- 10
- files supported in a single check
- 5
- supported upload formats: PDF, DOCX, TXT, RTF and ODT
- 2
- separate analyses: plagiarism checking and AI detection
- 150
- words available daily without an account
How we approach trust
Trust should come from clarity, not bigger claims.
Plagum avoids claims such as “100% accurate” or guarantees that an automated result can determine authorship. Plagiarism matches still need context, and AI indicators remain probabilistic.
We focus instead on clear report terminology, separate plagiarism and AI results, practical privacy controls and explanations of what each metric can — and cannot — tell you.
What the results do not mean
Three limits worth keeping clear.
Not a plagiarism verdict
A source match shows similarity. Quotations, citations, references and common wording can all produce legitimate matches.
Not proof of AI authorship
An AI result estimates writing patterns. It cannot prove who wrote a document or whether a specific person used AI.
Not a substitute for human review
Automated results should not be the sole basis for academic, disciplinary, employment or legal decisions.
