Core rule: AI may support research and learning, but it must not replace the author’s intellectual contribution, factual verification, independent judgment or accountability.
1. Purpose
This policy protects the originality, credibility and educational value of work submitted to, reviewed through, published on, or produced with services offered by ResearchBridge Africa. It explains how plagiarism and artificial-intelligence use will be prevented, disclosed, assessed and addressed.
ResearchBridge Africa supports responsible digital tools and does not treat all AI assistance as misconduct. The decisive questions are whether the use was authorised, transparently disclosed, independently verified, and consistent with the user’s institution, programme, assignment, discipline and applicable law.
2. Scope
This policy applies to students, authors, lecturers, supervisors, reviewers, editors, tutors, consultants, ambassadors, staff, contractors and administrators. It covers research topics, proposals, abstracts, dissertations, theses, articles, working papers, repository submissions, reviews, tutorials, assessments, research instruments, datasets, code, analyses, media, translations, support services, platform messages and uploaded evidence.
It also covers content created wholly or partly with generative AI, language models, paraphrasers, translators, coding assistants, image generators or similar systems. A stricter applicable rule imposed by an institution, journal, funder, ethics committee, supervisor or course prevails.
3. Key definitions
4. Plagiarism and related misconduct
Prohibited conduct includes copying words, ideas, arguments or structure without citation; patchwriting or close paraphrasing; uncited translations; submitting purchased, impersonated or another person’s work; prohibited duplicate submission; misappropriating confidential work, data, instruments, images, code or feedback; fabricating or concealing sources, data, ethics approval, results, authorship or contributors; and using AI or paraphrasing tools to disguise copied material or defeat originality checks.
Common knowledge normally needs no citation, but discipline-specific facts, distinctive explanations, data, theories, methods, quotations and derived arguments must be acknowledged in the appropriate academic style.
5. Similarity screening
ResearchBridge Africa may screen work before review, during an investigation, after revision or after publication. Checks may include text similarity, references, images, data and other integrity indicators.
No automatic percentage rule: A similarity percentage alone will not establish plagiarism or determine a penalty. Legitimate overlap may raise a score, while a low score does not prove originality.
An authorised human reviewer will interpret results in context, considering attribution, extent, intent, educational level, discipline, recurrence and effect on originality. The platform may request drafts, notes, source files, data, ethics records or an oral explanation.
6. Responsible use of AI
AI tools cannot be authors, reviewers or accountable contributors. A human author remains responsible for every statement, citation, calculation, interpretation, image, dataset and conclusion.
Permission is context-specific. A university, lecturer, journal, funder, ethics approval, client instruction or law may prohibit an otherwise generally permitted use.
7. Author and student duties
- Write and understand work submitted in your name and retain meaningful human control.
- Personally read and verify every cited source; AI output is not a source unless the research itself studies it.
- Independently check facts, calculations, quotations, references, code, translations, images and interpretations.
- Keep reasonable evidence of the research process, including drafts, notes, data, code versions and relevant prompts or outputs.
- Obtain permission before entering personal, confidential, copyrighted, unpublished, proprietary, sensitive or participant data into AI.
- Make and, where needed, update the required AI declaration.
8. AI-use disclosure
Users must disclose AI use that materially influenced content, method, analysis, coding, visual production, interpretation or presentation. Routine spelling and grammar checking that creates no substantive content may be exempt unless another rule requires disclosure.
The declaration should identify the tool and provider or version where known, date or period of use, task, affected section or output, human verification and revision, and whether protected data was entered.
Model declaration: “I used [tool/version] on [date/period] to assist with [specific task/sections]. I reviewed and revised the output, independently verified all facts and references, and accept full responsibility for the submitted work. [No personal, confidential or participant data was entered / approved safeguards used: …].”
9. Reviewers, editors and service providers
Unpublished submissions, reviews and applicant information are confidential. They must not be uploaded, in whole or identifiable part, to an external AI tool unless ResearchBridge Africa has expressly approved the tool and required notices or permissions are in place.
- Approved supporting AI use must be disclosed to the editor.
- AI must not decide acceptance, rejection, scoring, misconduct or sanctions.
- Human reviewers and editors remain responsible for accuracy, fairness, confidentiality, conflicts and final wording.
- Research support must guide, teach, edit or analyse within scope; it must not become ghostwriting, fabrication, impersonation or contract cheating.
10. AI-detection tools
AI-content indicators may form one part of a broader assessment, but an automated score is not conclusive proof. Before adverse action, an authorised person must examine the work and evidence, invite an explanation where appropriate, and record the reasons. A user will not be required to prove a negative solely because a detector produced a score.
11. Submission declarations and consent
The submitting author must confirm that the work is original to the extent claimed; sources, authors and contributors are identified; AI use is disclosed; data, permissions and ethics statements are genuine and lawful; there is no undisclosed conflicting submission or ownership claim; and proportionate integrity screening and record retention are accepted.
12. Reporting and preliminary assessment
Concerns may be reported through the platform’s complaint or academic-integrity channel with available evidence. Knowingly false, malicious or retaliatory allegations may lead to action. The Academic Integrity Officer or delegated editor will determine scope, immediate protective needs and whether a formal inquiry should begin. Anonymous reports may be considered where sufficient evidence exists.
13. Fair investigation procedure
- Preserve the submission, reports, versions, declarations and audit records.
- Conduct an initial human review and identify specific concerns and evidence.
- Notify the respondent clearly, subject to necessary confidentiality.
- Allow a response and supporting evidence, normally within 10 business days.
- Obtain independent advice where needed and manage conflicts.
- Decide on the balance of probabilities, recording evidence, reasons, severity and outcome.
- Communicate the decision, corrective action, appeal route and deadline.
Straightforward cases should normally conclude within 30 business days after the response. Complex cases may take longer, with reasonable updates.
14. Outcomes and proportionate action
Action will reflect seriousness, extent, intent, educational context, history, cooperation, harm and whether the record can be corrected.
Possible response: Guidance, correction, resubmission, warning or a required tutorial.
Possible response: Return or reject the work, suspend review or publication, issue a formal warning, require training or temporarily restrict features.
Possible response: Reject, remove or retract content; suspend or terminate access; cancel related certification or improper earnings; preserve evidence; and make a lawful, necessary referral.
Financial action follows the applicable Terms and refund or payout rules. An earned amount will not be withheld solely as punishment without an agreed rule or lawful basis, but fraudulent, undelivered or misconduct-tainted services may be declined, reversed or disputed through the proper process.
15. Corrections, removal and post-publication action
The platform may add a notice, correct metadata, replace a version, restrict access, remove content or issue a withdrawal or retraction statement. The action must describe the record accurately without overstating the finding. Material findings may be shared with relevant bodies only for a legitimate, proportionate and lawful reason.
16. Appeals
A respondent may appeal within 15 business days for a material procedural error, previously unavailable relevant evidence, a clearly unreasonable interpretation, conflict of interest or disproportionate outcome. A person or panel not materially involved in the original decision will review the appeal. The platform appeal outcome is final without limiting external rights.
17. Privacy, security and records
Integrity processing must follow the Privacy Policy, applicable data-protection duties and role-based access controls. Sensitive evidence is restricted to authorised persons. Records may include submissions, screening reports, declarations, relevant prompts, correspondence, decisions and audit entries, retained only for a documented or legally necessary period.
18. Training and prevention
ResearchBridge Africa will provide or signpost guidance on citation, paraphrasing, ethics, source verification, responsible AI, privacy and authorship. Educational correction is preferred for appropriate first-time minor mistakes; deliberate or serious misconduct receives stronger action.
19. Governance and review
The Super Admin will designate responsibility for screening, investigations, appeals and policy review. Role separation should prevent the same person from reporting, investigating and deciding an appeal where practicable. The policy will be reviewed at least annually and after significant changes in AI, academic practice, platform services, law or guidance.
20. Relationship with other rules
This policy should be read with the Terms, Privacy Policy, Peer-Review Policy, Publication and Copyright Policy, Research Ethics Policy, Reviewer Code of Conduct, Complaints Procedure, and Refund and Payout Policy. Where provisions conflict, the more specific applicable rule governs unless law requires otherwise.
21. Guidance basis
This policy is informed by recognised principles from the Committee on Publication Ethics (COPE), the International Committee of Medical Journal Editors (ICMJE), and UNESCO’s guidance for generative AI in education and research. These materials guide good practice but do not replace applicable law, institutional rules or professional advice.
Appendix A — Quick decision guide
Appendix B — Submission declaration
I confirm that this submission is my/our work to the extent stated; all sources and contributors are properly acknowledged; all material AI use has been disclosed; I/we have verified the accuracy of the content and references; and I/we accept responsibility for the submission. I/we understand that ResearchBridge Africa may conduct proportionate academic-integrity screening and apply its fair-review procedure.
This policy should receive final review from ResearchBridge Africa’s academic advisory team and a qualified Ghanaian legal and data-protection adviser before the wider public launch.