// 04 AI AND PLAGIARISM CHECKER

Detect AI-generated content and plagiarism in learner submissions

Every submission checked. Assessors get a clear report before they open the file, so decisions are made on evidence, not instinct.

Request a callback

What it is

Pearl's AI and Plagiarism Checker scans learner submissions for two distinct issues: content that has been copied from external sources (plagiarism), and content that shows markers of AI generation. It returns a structured report for each submission that assessors and quality teams can use to make informed academic integrity decisions.

The checker runs before assessment, so by the time an assessor opens a submission they already know whether it has flagged any concerns. It does not make a ruling on whether a learner has breached integrity policies, that is an assessor and provider decision. It surfaces evidence clearly so that decision can be made consistently and fairly.

The tool is designed for the specific characteristics of FE and skills-sector learner writing, not for academic prose at degree level, which means the detection model is calibrated to the kind of submissions your assessors receive.

Who it is for

The tool is used by FE colleges, ITPs, and HE providers managing learner assessments at volume. It is particularly relevant for providers where AI use in learner submissions has become a quality concern, where assessors have inconsistent approaches to handling suspected plagiarism, or where the volume of submissions makes manual checks impractical.

Quality leads and IQAs use the aggregated reports to identify patterns across cohorts or programmes. Assessors use submission-level reports to inform their marking decisions and, where necessary, to initiate formal academic integrity processes. It integrates directly with SAM so both tools run in parallel on each submission.

Key features

AI detection

Identifies linguistic and structural patterns associated with AI-generated text, with a confidence score and highlighted passages.

Plagiarism detection

Checks submissions against a broad source database including academic publications, web content, and previously submitted work within your institution.

Submission-level report

Assessors receive a clear, readable report for each submission before they begin marking.

Cohort and programme reporting

Quality leads can view aggregate AI and plagiarism rates across cohorts, flagging programmes where patterns are emerging.

Evidence export

Reports are exportable for use in academic integrity hearings or Ofsted evidence packs.

SAM integration

Runs automatically alongside SAM so every SAM-processed submission is also checked.

Proof points

Calibrated to FE and skills-sector learner writing, not academic prose, reducing false positive rates for vocational submissions.
Reports are available within minutes of submission upload, before assessors begin marking.
Used by providers managing thousands of learner submissions per month across multiple programmes.

How it works

Submissions are uploaded to the checker either manually or automatically via integration with Atom or SAM. The checker processes each file and returns a report covering AI likelihood score, plagiarism match percentage, matched sources, and highlighted passages of concern.

Assessors log in to a submission queue where reports are displayed alongside the original file. They can annotate the report, flag for IQA review, or proceed to marking with the report attached to the submission record.

Configuration involves setting your institution's threshold levels for when submissions are automatically flagged for review, and confirming the source database scope relevant to your programmes. Most configurations are live within a day.

Integrations & compatibility

Pearl Atom LMS (automatic submission routing) Pearl SAM (parallel processing) File upload for standalone use Export to PDF for evidence packs

FAQs

Does the tool decide if a learner has cheated?

No. The tool returns evidence, a likelihood score, matched passages, and source links. The decision on whether a learner has breached your academic integrity policy is always made by your staff.

How accurate is the AI detection?

The tool is trained on data relevant to FE and skills-sector submissions and is updated as AI writing tools evolve. No detector is 100% accurate, and Pearl recommends treating AI detection scores as evidence to inform a human decision, not as a definitive ruling.

Can it check submissions that were written in a language other than English?

The tool currently supports English-language submissions. Contact Pearl if multilingual detection is a requirement for your provision.

Will this flag learners who have used AI tools legitimately, such as grammar checkers?

The tool distinguishes between light AI-assisted editing and substantively AI-generated content. Short passages or corrected text typically do not generate a high AI likelihood score. Your institution's AI use policy will determine what level of use is acceptable.

How does it handle previously submitted work from other institutions?

The plagiarism check includes a database of previously processed submissions from across the Pearl network, in addition to external sources. Submissions that match work submitted elsewhere in the network will be flagged.