Academic integrity

A tool that helps you do the work — not one that pretends you did

PassResearch is designed so that unsupported claims are visible rather than convincing. This page states plainly what the platform will and will not do.

A quiet study area with desks, shelving and natural light.
Work that stands on its own record.
01

Principles

Five rules the platform is built around

These are constraints on the product, not aspirations. Where a feature could break one of them, the feature is not built.

01

Evidence before eloquence

Every claim that needs support is expected to point to a source record. When the assistant cannot find relevant evidence, it says so and stops rather than writing something plausible.

02

Nothing is invented

No fabricated DOI, author, year, journal, volume, page range, quotation, participant response, survey finding or statistical result. Missing information is reported as missing.

03

States are never conflated

A discovered record is not a verified one. An abstract is not a full text. An AI suggestion is not an approved revision. The interface keeps these visibly separate.

04

Your work stays yours

Your private research content is treated as confidential and is not shared with other users. Uploaded files are served only through authenticated routes that check ownership.

05

You stay responsible

Responsibility for the accuracy of sources, the appropriateness of methods, ethical approval and compliance with your institution remains with you as the researcher.

02

AI assistance

What the assistant is allowed to do

Grounded generation is a constraint, not a marketing word. These rules are enforced in the server path, not merely described here.

  1. 01Research assistance retrieves the source records it used before generating text that needs citations.
  2. 02Citations suggested by the assistant are filtered against the records actually retrieved — they cannot reference something that was not found.
  3. 03Generated text is presented as a suggestion to accept, edit, reject or regenerate. Approved sections are never silently overwritten.
  4. 04No numerical confidence scores are invented. Verification labels are shown only where there is a defensible basis.
  5. 05If there is not enough context — no data, no sources, no methodology — the assistant asks for it or states the limitation.
03

Privacy & data

How your research data is handled

Your project is your work. The defaults keep it close, and every external step is something you choose and can see.

A student working on their own laptop at a study desk.

Privacy & data

Local-first by default

Running locally, your database and uploads live in the project folder on your computer. Nothing leaves the machine unless you configure an external provider and use it.

Two people comparing printed documents page by page.

Privacy & data

Minimum necessary sharing

Where an external service is used — a scholarly provider or an AI model — only the query or excerpt needed for that step is sent, not your entire dissertation.

Researchers working with equipment in a teaching laboratory.

Privacy & data

Sensitive participant data

When research involves human participants or sensitive information, PassResearch guides you toward your institution's ethics process and keeps datasets private to your account.

What PassResearch will not claim

It will not promise plagiarism-free work, undetectable AI use, a guaranteed grade, or automatic compliance with any university’s official requirements. Institutional templates are used only when they have actually been obtained and reviewed; otherwise your project is marked as student-configured.

Start honestly

Build a project you can defend

Every source you save keeps its provenance, every figure is counted from your own records, and every suggestion stays a suggestion until you accept it.