Generative AI Policy

Generative AI and Automated Tools Policy

Applies to: all 6 EP Journals Group journals.
Effective: 1 May 2026, version 1.0.

This policy follows the DOAJ best-practice guidance on AI and automated tools and the COPE position on Authorship and AI Tools. It covers what authors, reviewers, and editors at Journal of Natural Science Research and Review may and may not do with generative AI tools (large-language models, image generators, code generators, automated paraphrasers, AI-assisted research tools).

For authors

What you must disclose

If you used a generative AI tool in your research or in preparing the manuscript beyond straightforward language correction, editing, or formatting, you must disclose this. The disclosure goes in two places:

  1. In the Methods section if AI was used as part of the research itself (data analysis, code generation, image generation, simulation, retrieval augmentation), with enough detail that a reader can understand what role AI played.
  2. In the Acknowledgements if AI was used to assist with writing, structuring, or editing the manuscript text in a substantive way.

The disclosure should name the tool (e.g., "ChatGPT GPT-4o, OpenAI, accessed June 2026"), describe what you did with it, and confirm that you take full responsibility for the output.

We do not require disclosure for the kinds of automation that are now routine in scholarly writing: spell-check, grammar-check (e.g., Grammarly's basic features), reference manager autocompletion, citation lookup tools, or translation tools used for sentence-level translation that the author then reviewed.

What is not allowed

  • AI cannot be listed as an author. Authorship requires accountability for the work and the ability to defend it. AI tools cannot meet that bar. This is the COPE position and we apply it strictly.
  • AI cannot be cited as a source for facts or arguments. AI outputs are not retrievable, are not stable, and are not authored in any meaningful sense. If you use an AI to summarize a body of literature, the citations in your paper must be to the original sources, not to the AI's response.
  • You are responsible for verifying AI output. AI tools confidently produce false references, false statistics, false historical facts, and plausible-sounding nonsense. If you used an AI to find references, summarize papers, or draft sections, you must verify the output against primary sources before submission. We have rejected submissions where reference-list checking revealed citations that don't exist.
  • AI cannot be used to fabricate data or images. Generating synthetic data and presenting it as observed data is research fraud. Generating images that purport to show actual experimental results is research fraud. Both lead to retraction and may be reported to the authors' institution.

Practical guidance

Using an AI to brainstorm an outline, smooth out a difficult paragraph, suggest alternative phrasing, or check whether a sentence is clear is fine and does not require disclosure beyond ordinary acknowledgement of editorial assistance. Using an AI to write whole sections of a paper, generate a literature review, or produce data analysis without independent verification is misconduct.

If you are unsure whether a particular use should be disclosed, disclose it. The cost of unnecessary disclosure is zero; the cost of missing disclosure that surfaces later can be retraction.

For peer reviewers

What you must not do

  • Do not paste a manuscript into a generative AI tool. Manuscripts under review are confidential. Pasting them into ChatGPT, Claude, Gemini, or any other commercial AI service breaches confidentiality and may constitute a leak of unpublished research. We treat this as a serious ethical breach.
  • Do not use AI to draft your review. AI-generated reviews are typically generic, miss the specific scholarly contribution, hallucinate references, and tend to converge on similar phrasing across reviewers. A peer review's value is in your specific expertise, which AI cannot supply.
  • Do not use AI to summarize a manuscript for your own convenience. Same confidentiality concern. Read the manuscript yourself; that is what you agreed to do when you accepted the review invitation.

What is acceptable

You may use ordinary writing tools — grammar check, spell check — to polish your review's language after you have written it. You may not use AI to generate the substance of the review.

If you suspect AI was used in a manuscript you are reviewing (overly generic phrasing, fabricated-seeming references, methodology described at a level that doesn't match the author's claimed familiarity), flag this confidentially to the editor.

For editors

The Editor-in-Chief and Section Editors do not use generative AI to:

  • compose decision letters
  • summarize manuscripts for editorial committee discussion
  • generate reviews of manuscripts where a reviewer dropped out
  • write editorial commentary, editorials, or other journal content presented as editorial voice

Editors may use AI tools that are part of the editorial workflow if the tool's role is disclosed and verified by a human:

  • iThenticate's AI-assisted content detection features (we use these and disclose this here)
  • automated language quality screening at desk-screening stage
  • automated reviewer-suggestion tools (the editor reviews and chooses)

In every case, a human editor verifies the output before any decision is made.

What we do at the publisher level

EP Journals Group uses the following automated tools across all six journals, and we disclose them here:

  • iThenticate (Crossref Similarity Check) for plagiarism screening at desk-screening
  • Crossref's reference-matching service for citation linking
  • OJS automated similarity flags to draw editor attention to high-similarity submissions

We do not use AI to evaluate the scholarly quality of submissions, to draft editorial decisions, or to generate any content presented as the journal's voice (editorials, calls for papers, announcements).

We monitor developments in AI-content detection and may add other automated tools to the workflow. Any addition will be disclosed in this policy and reviewed by the Editorial Board before adoption.

When this policy was last updated

Effective: 1 May 2026 (v1.0).

This is a relatively new policy area and we expect to revise it as norms develop. Updates are announced 30 days before they take effect, applied prospectively to new submissions, and have no retroactive impact on already-published articles.

Editorial contact: editor@jnsrr.ep-journals.org
Ethics contact: ethics@ep-journals.org
External references: DOAJ AI guidance · COPE Authorship and AI Tools · WAME recommendations on chatbots