Artificial Intelligence Policy

Quick Reference Guide

Stakeholder	Permitted AI Use	Prohibited AI Use	Disclosure Required
Authors	Grammar/spell check, literature search assistance, translation support, formatting, explanatory diagrams, data visualisations from real data	Generating research content, generating or altering data (except disclosed synthetic data, see 2.2), creating or altering primary research images, graphical abstracts from generative image tools, generating references without verification	Yes – for any use beyond basic grammar, spelling, and punctuation correction (see 2.4)
Reviewers	Grammar check of own reports (offline tools only)	Uploading manuscripts to AI tools, generating review content, evaluating research	Yes – any use beyond basic spelling and grammar checks, disclosed to the editor
Editors	Plagiarism detection, reviewer matching, technical screening, offline grammar check of own correspondence	Making acceptance/rejection decisions, generating decision letters, uploading confidential manuscripts	N/A – journal discloses approved tools publicly

1. Introduction

This policy governs AI use in manuscript preparation, peer review, and editorial processes for the West African Journal of Open and Flexible Learning. AI may assist but never replace human judgement, expertise, or accountability. Authors remain responsible for all content.

This policy aligns with international standards established by the Committee on Publication Ethics (COPE), the International Committee of Medical Journal Editors (ICMJE), and the World Association of Medical Editors (WAME).

Definitions:

Generative AI: creates original content (ChatGPT, Claude, DALL-E)
AI-assisted tools: support tasks without generating content (Grammarly, reference managers)

Basic spell-checkers require no disclosure.

2. Policy for Authors

2.1 What Authors May Use AI For

Without disclosure:

Basic grammar, spelling, and punctuation correction
Reference formatting and citation management
Looking up individual words, phrases, or sentences with standard translation tools
AI-powered plagiarism checkers on your own work before submission
AI features within assistive technology used solely for accessibility purposes

With mandatory disclosure:

Revisions beyond basic grammar/spelling, such as rewriting paragraphs, restructuring sections, or rephrasing arguments for clarity
Assisting literature searches or generating summaries (verify all sources independently by reading original articles)
Using AI to translate substantial portions of a manuscript (the translation must be verified by a language expert and a subject-matter expert)
Creating data visualisations (plots, charts, graphs) with AI-assisted tools, provided the output comes directly from your real underlying data and the method is reproducible and described in the Methods section
Creating explanatory images such as flow charts, decision trees, timelines, or schematic diagrams, with the tool, version, and use stated in the figure caption and in your AI disclosure
Any use in research methodology, data collection, or analysis
AI-assisted statistical software or code generation (e.g., R with AI code completion)

2.2 What Authors Must Never Use AI For

The following constitute research misconduct:

Generating original research content: AI cannot write hypotheses, methods, results, discussions, or conclusions
Fabricating or manipulating data: AI must never create, alter, or enhance research data or experimental results
- Exception for synthetic data: Artificially generated data may be used only where creating it is part of the research design, for example to train or test predictive educational models, to test learning management system (LMS) algorithms, or to produce privacy-preserving “data twins” that protect student identity. The method must be described reproducibly in the Methods section (tool, version, developer, and how the data were generated), and the data must be labelled as synthetic wherever they appear. Presenting synthetic data as real student performance or classroom outcomes is research misconduct.
Creating or altering primary research images: Generative AI must not be used to create or alter images that represent observed or experimental data (for example photographs, micrographs, scans, or blots). Adjustments such as brightness or contrast must be made with established image-processing software
- Exception: AI tools that form part of the research method (e.g., AI-assisted medical image analysis, machine learning models for data interpretation) are permitted when fully described in the Methods section, including tool name, version, and developer. They are not treated as image fabrication.
Creating graphical abstracts with general-purpose generative AI image tools. Use dedicated scientific illustration tools instead
Generating images that reproduce copyrighted artwork, real people, or identifiable brands
Generating references: AI-generated citations must be independently verified; fabricated references constitute misconduct
Undisclosed content generation: Failing to disclose AI use in substantive text generation violates transparency requirements

2.3 Responsible Use of AI Tools

Before using any AI tool, check its terms and privacy policy. Your unpublished manuscript, data, and any personal information must not be retained by the tool or used to train it, and the tool must not restrict how you publish its outputs. Check outputs for factual errors and bias.

2.4 How to Disclose AI Use

Where to disclose:

AI declaration: Include a short statement titled “Declaration of AI use” before the references if AI assisted with manuscript preparation or writing. If your article is published, this statement is published with it.
Methods section: If AI was used in research design, data collection, analysis, or figure creation, describe it here in enough detail to be reproduced.

Required information:

Tool name, version, and developer (e.g., “ChatGPT, GPT-4o, OpenAI”)
Specific purpose (e.g., “restructuring Introduction for clarity”)
Date(s) of use
Confirmation that you reviewed and edited the output and take responsibility for the content

Example:

“The authors used ChatGPT (GPT-4o, OpenAI) on 15 March 2026 to restructure sections of the manuscript for clarity. All AI-generated text was reviewed, fact-checked, and rewritten by the authors to reflect their own analysis. The authors take full responsibility for all content.”

2.5 Authorship Rules

AI tools cannot be listed as authors. Only humans who meet ICMJE criteria (making substantial intellectual contributions, approving the final manuscript, and accepting accountability) may be authors.

2.6 Examples

Scenario	Acceptable?	Why?
Using AI to fix grammar	✓ Yes (no disclosure)	Standard editing function
Using AI for literature summary, then reading sources and writing your own analysis	✓ Yes (disclose in AI declaration)	Human verification and synthesis
Using AI to write your entire Discussion	✗ No	Violates authorship accountability
Using AI to generate citations without checking	✗ No	Reference fabrication
Using AI to create a graphical abstract	✗ No	General-purpose generative image tools are not permitted for graphical abstracts
Using AI-assisted imaging software as part of research methods	✓ Yes (disclose in Methods)	Legitimate research methodology
Using R with AI code completion (e.g., GitHub Copilot)	✓ Yes (disclose in Methods)	AI-assisted tool for analysis
Using AI to help draw a flow chart or timeline	✓ Yes (disclose in caption and AI declaration)	Explanatory image, not research data
Using AI-assisted code to plot your real data	✓ Yes (disclose in Methods)	Visualisation derived directly from underlying data
Using AI to generate or enhance a micrograph or scan	✗ No	Alters primary research data

3. Policy for Peer Reviewers

3.1 Confidentiality is Paramount

Reviewers must never upload manuscripts to external AI systems. Most AI tools retain inputs and may use them in training data, breaching confidentiality. This applies to manuscript text, figures, tables, and your own review reports.

3.2 Limited Permitted Uses

Reviewers may only use AI for:

Grammar-checking their own reports using tools that run on their own device and do not transmit, retain, or learn from the content
Bibliographic searches to verify claims
Checking statistics with conventional (non-AI) software that runs securely on their own device, using only the data supplied with the manuscript

Basic spelling and grammar checks need no disclosure. Disclose any other AI use to the editor in confidential comments.

3.3 Prohibited Uses

Reviewers cannot use AI to generate review content, assess scientific validity, or evaluate findings. Peer review requires expert judgement that AI cannot replicate. Reviewers should not use AI to verify authors’ statistical claims, as AI may provide inaccurate assessments; raise concerns with the editor instead.

This does not prevent reviewers from running their own calculations in conventional statistical software. If you have concerns about the statistical methods, raise them with the editor.

4. Policy for Editors

4.1 Permitted and Prohibited Uses

Editors may use journal-approved AI for:

Plagiarism detection and manuscript screening
Identifying potential reviewers based on expertise
Technical editing post-acceptance (formatting, copyediting)
Grammar-checking their own correspondence using offline tools that do not transmit or retain content

Editors must never use AI to:

Make acceptance/rejection decisions
Write decision letters or author correspondence
Conduct scientific evaluation of manuscripts
Upload manuscripts to external systems

Editorial independence requires human judgement.

4.2 Enforcement

Editors must:

Review all author AI disclosures for compliance
Investigate suspected violations using COPE guidelines
Document investigations and maintain records
Consult with the editorial board on complex cases

5. Transparency and Enforcement

5.1 Journal Transparency

This journal publicly discloses AI tools used in editorial workflows (plagiarism detection, manuscript screening) and reviews this policy annually to reflect technological developments and evolving international standards.

5.2 Violations and Consequences

Severity is assessed using the same tiers as the Publication Ethics and Malpractice Statement.

Severity	Examples	Consequences
Minor	Incomplete disclosure of permitted AI use	Request for corrected disclosure or revision
Moderate	Undisclosed AI generation of substantive text; undisclosed AI use in methods or figures	Manuscript rejection or publication delay
Serious	AI-fabricated data, references, or images; presenting synthetic data as real; plagiarism; uploading confidential manuscripts to AI tools; deliberate or repeated violations	Authors: Rejection or retraction, institutional notification, submission ban Reviewers and editors: Removal from their role and notification of their institution

Investigations follow COPE guidelines. Authors may appeal editorial decisions through the process in Section 8.2 of the Publication Ethics and Malpractice Statement. Reviewers, editors, and others may raise complaints under Section 8.1.

5.3 Detection Methods

The journal uses mandatory disclosure statements in submission forms, editorial review of AI disclosures, AI detection tools where appropriate, and image forensics software.

AI detection tools are unreliable and can wrongly flag human-written text, particularly writing by non-native English speakers. Detector results are never used on their own as grounds for rejection or any other action. They may only prompt a closer editorial review, and any decision must rest on additional evidence and give the author the opportunity to respond.

6. Conclusion and Policy Review

This policy ensures AI use remains transparent, ethical, and human-led. We review this policy annually to reflect technological advances, evolving international standards (COPE, ICMJE, WAME), and legal developments in copyright and data protection. We welcome feedback from authors, reviewers, and readers to strengthen this living policy.

Read this policy alongside our Publication Ethics and Malpractice Statement, our Peer Review Process, and our Author Guidelines.

Frequently Asked Questions

Q: I used AI to rephrase a sentence. Must I disclose this?

A: It depends on what the AI changed. Basic corrections to grammar, spelling, and punctuation need no disclosure. If the AI changed the wording, structure, or argument of the text, even in a single sentence, disclose it in your AI declaration. If you are unsure, disclose.

Q: Can I use AI to translate my manuscript from another language into English?

A: Yes, but have a language expert and a subject-matter expert verify the translation’s accuracy. Looking up individual words or sentences needs no disclosure, but disclose substantial translation work in your AI declaration.

Q: What if I’m unsure whether to disclose?

A: Default to disclosure. Transparency cannot harm your submission; non-disclosure of substantive AI use may result in rejection.

Q: Can I use AI to check my own manuscript for plagiarism before submission?

A: Yes. Using AI plagiarism-detection tools on your own work is acceptable and recommended. You may mention this briefly in your AI declaration if relevant.

Q: Can I use AI to format my references into different reference styles?

A: Reference managers such as Zotero, Mendeley, or EndNote are fine. If you use a generative AI tool to reformat your references, you do not need to disclose this, but you must check every reference against the original source afterwards. These tools can drop references, mix up authors, titles, and DOIs or URLs, or invent details. You remain responsible for the accuracy of your reference list.

Q: What if I discover fabricated AI references after submission?

A: Contact the editorial office immediately. Honest mistakes are correctable; deliberate fabrication is misconduct.

Q: Can I use AI to help design my study or develop hypotheses?

A: AI may support brainstorming, but you must develop and own the final research design. Disclose AI’s role in the Methods section if it substantively influenced your approach.

Q: Can I use AI-powered statistical software or code completion tools like GitHub Copilot?

A: Yes. Disclose this use in the Methods section (e.g., “Statistical analysis conducted using R 4.3.1 with GitHub Copilot assistance for code generation. All code was verified by the authors.”).

Q: Can reviewers use AI to verify statistical claims in manuscripts?

A: No. AI may provide inaccurate statistical assessments. You may run your own calculations in conventional statistical software. If you have concerns about authors’ statistical methods, raise these with the editor rather than relying on AI verification.

Q: What counts as “substantial revision” that requires disclosure?

A: Revisions beyond basic grammar, such as rewriting entire paragraphs, restructuring sections, or rephrasing complex arguments. When in doubt, disclose.