About | Plagiarism Checker

PlagPointer offers Plagiarism and AI detection that goes much further than a similarity score

PlagiarismChecker.net has been online since 2009. Today, our checker is PlagPointer: a British-developed plagiarism and AI detection service built for students, educators, writers, publishers, businesses and developers.

One scan can uncover copied text, disguised paraphrasing, translated plagiarism, AI-generated writing, suspicious text manipulation and copied source code – with detailed evidence in a downloadable report.

What does PlagPointer search?

60 trillion+ web and search-engine sources
16,000+ open-access journals and publications
1 million+ shared/internal documents available for database matching
20+ source-code repositories
100+ languages for plagiarism detection
30 languages for AI detection

The academic sources include publications from major names such as Elsevier, Wiley and Frontiers.

A lot more than copy-and-paste detection

PlagPointer can be used as a straightforward checker with the standard settings, or you can open the advanced controls and decide exactly how the scan should work.

Exact plagiarism – finds passages that match a source word for word.
Changed wording – detects smaller alterations rather than relying only on identical sentences.
Paraphrased plagiarism – looks for text that has been rewritten while retaining the meaning of the source. The detection technology distinguishes identical matches, minor changes and related/paraphrased wording.
AI-generated text – checks eligible documents for writing associated with major generative-AI systems. The detection technology covers model families including GPT/OpenAI, Gemini, Claude, DeepSeek, Llama and others, and is continually updated as models change.
A score you can actually explain. PlagPointer doesn’t just hand you a number. Explainable AI Logic shows why a passage was flagged – the phrasing patterns statistically associated with AI writing, and matches against AI-generated text already published elsewhere. When you have to defend a result, “the software said so” is not an argument – but our evidence is.
Adjustable AI detection sensitivity – choose Extra Safe, Balanced or Extra Sensitive depending on how cautiously you want the AI analysis to behave. Balanced is the recommended default. Extra Safe is designed to minimise false positives, while Extra Sensitive is more aggressive when looking for AI text that has been heavily edited, rewritten or passed through a “humaniser”.
AI and plagiarism together – you do not need to submit your work to two separate services. A normal PlagPointer document scan can check both.
Cross-language plagiarism – check whether English text may have originated from material written in another language. Cross-language matching supports 30+ target languages, including Dutch, French, German, Spanish, Italian, Portuguese, Polish, Greek, Hindi, Russian, Ukrainian and many more.
Source-code plagiarism – upload actual code rather than converting it to plain text first. Supported formats include Python, PHP, JavaScript, TypeScript, Java, C, C++, C#, Go, Ruby, Swift, Scala, CSS, shell scripts and more. Code can be compared against public sources and a search index that includes 20+ code repositories.
Text-manipulation detection – look for attempts to interfere with a plagiarism check using hidden text, invisible characters, substituted lookalike characters and other forms of document manipulation.
Search only the websites you want – restrict a plagiarism check to one or more specific domains.
Exclude websites you do not want – useful if, for example, your own website is dominating the results or you already know about a particular source. Domain inclusion and exclusion are supported directly by the plagiarism engine.
Exclude material that should not count – remove quotations, citations, references and tables of contents from matching. PlagPointer can also ignore section headings, repeated page/header/footer material and source-code comments.
Choose how deeply to search – Low, Medium and High sensitivity settings allow you to control how thoroughly the scan looks for smaller matches.
Optional shared-database matching – comparison against shared/internal submissions is available when you want it, but it is off by default in PlagPointer.
Detailed PDF reports – see the matching material rather than being given a mysterious percentage with no context. Reports include the evidence behind the result and remain available in My scans for later download.
No subscription required – buy credits when you need them. One standard credit covers up to 250 words, and PlagPointer shows the required credit cost before you start the scan. Credits do not expire. Cross-language checking costs extra because each additional language requires another search. The underlying API likewise defines one billing page as 250 words.

AI detection with numbers behind it

PlagPointer does not rely on a detector whose accuracy is simply described as “good”.

The detection technology behind PlagPointer publishes its own large-scale test results, and has also been assessed by independent researchers. In the developer’s V10 testing at the default Balanced sensitivity:

99.21% of AI-written texts correctly identified
99.97% of human-written texts correctly identified
0.026% false-positive rate

And independently verified. In peer-reviewed testing published in 2026, the engine behind PlagPointer scored 99.6 out of 100 for detecting AI-generated scientific articles – significantly ahead of every other detector tested bar one (Kaya et al., International Journal of Oral and Maxillofacial Surgery, 2026). A separate 2026 study in the Journal of Advances in Information Technology tested nine detectors against paraphrased and deliberately disguised AI text, where most collapsed: Turnitin fell to 45.7% and Grammarly to 19%. Ours was one of only three that held up.

We publish the limitations too: detection is less reliable on very short text, heavily hand-edited text, and hybrid human/AI drafts. That’s why we let you choose your sensitivity – and why we don’t pretend a score settles anything. In the same V10 testing mentioned above, the Extra Safe model reduced the reported false-positive rate to 0.009%, while the Extra Sensitive model increased detection of modified AI text with a reported false-positive rate of 0.05%. Increasing sensitivity therefore involves a trade-off: more aggressive detection can identify more altered AI material, but also increases the risk of a human-written text being flagged.

The evaluation was not based on a handful of ChatGPT answers. One Data Science test used 500,000 English texts – 300,000 human-written and 200,000 AI-generated – while a separate QA evaluation tested more than 220,000 human texts and 18,000+ AI-generated texts.

The AI test material included output from GPT-5 and other OpenAI models, Gemini, Claude and Grok. Importantly, the evaluation data was kept separate from the material used to train the detector.

Find plagiarism that has been translated

PlagPointer’s cross-language search is designed to find possible matches where material has been copied from a source in another language and translated before reuse.

For example, an English essay can be checked for possible matches with Dutch, German, French, Spanish, Polish or other supported sources.

This is a separate, deeper search and therefore uses additional credits. PlagPointer shows the increased cost before you start it.

Plagiarism checking isn’t just for essays

Check source code

A copied programming assignment may bear very little resemblance to an ordinary essay, but that does not make it impossible to check.

PlagPointer accepts real source-code files including .py, .php, .js, .ts, .java, .cpp, .cs, .go, .rb, .swift, .scala, .css, .sh and many more.

That makes it useful for:

programming coursework
coding assessments
outsourced development
submitted code
contributed code
checking whether code appears elsewhere

Source-code scans use normal plagiarism credits. AI-text detection is not applied to source-code files.

Look for attempts to hide plagiarism

Some plagiarism is harder to see because the document itself has been manipulated.

PlagPointer’s optional text-manipulation detection can look for techniques intended to confuse automated checking – for example replacing ordinary characters with visually similar Unicode characters or inserting material that is invisible to the reader.

It gives educators and reviewers another layer of information when a document does not look quite right.

Search the right sources – not just more sources

A large search index is useful. Being able to control it is better.

If you are a publisher checking an article that already appears on your own website, you can exclude your domain.
If you want to know whether content came from a particular group of sites, you can restrict the search to those domains.
If references and quotations are creating legitimate matches, exclude them.
If you are checking code, you can ignore comments.
If you want shared-database matching, turn it on. If you do not, leave it off.

You are not locked into one generic scan configuration.

Your evidence stays with the report

Every completed PlagPointer scan is saved to your account under My scans.

The PDF is there when you need to return to it – whether that’s five minutes later or weeks afterwards – along with the scan reference and date.

There is no need to repeat a scan just because you forgot to download the report.

Built in Britain

PlagPointer is developed and operated through PlagiarismChecker.net, a British website established in 2009.

The site began as a very simple free plagiarism checker and a resource about plagiarism, academic integrity and plagiarism-checking technology. Since those humble beginnings, we developed PlagPointer to bring those subjects together in the checker itself: sophisticated detection underneath, but without forcing ordinary users through an enterprise platform to use it.

Pay for the checking you actually need

There is no monthly subscription.

PlagPointer credits are sold in bundles and do not expire. A standard credit covers up to 250 words, so documents are charged in relatively small increments rather than large blocks.

Before the scan begins, you see the exact credit requirement.

Upload a document → choose your checks → see the price → start the scan.