Opens in a new tab

Home / Resources / AI in Research

How to Disclose AI Use in a Research Paper:
Journal Policies and Example Wording

Knowing how to disclose AI use in research comes down to three facts: which tool was used, for what task, and how its output was checked, reported as the receiving journal's policy requires. A disclosure is not an admission that the analysis is flawed; it is a record of how the work was produced. Journal requirements differ, especially between language editing, writing assistance and analysis-related tasks (International Committee of Medical Journal Editors [ICMJE], 2026; Zielinski et al., 2024).

AI in Research
WHO ASKSJournals, editors
WHEREMethods, declaration
KEEPInternal AI-use record
NEVERClaim unperformed checks
LAST CHECKEDNovember 2026

AI can support researchers in clarifying language, drafting analysis code and working through reporting questions. A transparent statement lets the researcher benefit from those tools while preserving responsibility for the submitted work.

How to disclose AI use in research: what journals require

The ICMJE recommends that authors disclose use of AI-assisted technologies in both the cover letter and submitted work, in the appropriate section. It also states that AI systems should not be listed as authors, because authors must take responsibility for accuracy, integrity and originality (ICMJE, 2026). The World Association of Medical Editors (WAME) has likewise discussed transparency and accountability in AI-assisted publication (Zielinski et al., 2024).

These are influential recommendations, not a universal rule for every journal or discipline. A journal may require a dedicated declaration, a Methods description, an acknowledgment, submission-system responses, or some combination. Check its current author instructions before submission and again after any policy update.

POLICY TABLE · Last checked November 2026 · rechecked every six months

What AI contributed to What to explain Likely location to check
Grammar, clarity or translation Tool, scope of editing, and author review AI declaration or acknowledgments
Drafting statistical code Tool, purpose, execution environment, verification and revisions Methods/statistical analysis and AI declaration
Analyzing data or preparing figures Which calculations were AI-assisted, how they were validated, and how inputs were protected Methods and AI declaration
Screening or extracting studies Which review stages used AI, human checking and disagreements Systematic-review Methods and AI declaration

A paper can legitimately mention AI in more than one place: a concise disclosure for transparency and a technical account in Methods when the tool affected the research procedure.

AI disclosure statement examples you can adapt

These are illustrative wording examples, not quotations from journals or statements that can be used without verification. Replace placeholders and remove tasks that did not occur.

Example 1: Language editing only

EXAMPLE WORDING (ADAPT TO THE JOURNAL)

During manuscript preparation, the authors used [tool and version, if available] to suggest improvements in grammar and clarity. All suggested changes were reviewed and edited by the authors, who take responsibility for the final manuscript.

Example 2: AI-assisted statistical programming

EXAMPLE WORDING (ADAPT TO THE JOURNAL)

[Tool and version] was used to suggest and troubleshoot [R/Python/SPSS] syntax for the prespecified analyses. The scripts were executed in [approved environment] using the study dataset. A qualified researcher/statistician inspected the code, checked analysis assumptions and compared selected results with independent calculations. The authors take responsibility for the methods, findings and interpretation.

Example 3: Systematic review screening

EXAMPLE WORDING (ADAPT TO THE JOURNAL)

[Tool and version] supported [title/abstract screening or data extraction] under predefined criteria. The review team independently checked [records/data fields], resolved discrepancies using [procedure], and reported the automation in the review workflow. Final eligibility and extraction decisions were made by the research team.

Do not write “independently checked” unless that check really occurred. Replace broad phrases such as “validated by experts” with verifiable steps, where possible. Avoid publishing confidential prompts, participant details or protected study documents merely to make the declaration longer.

What records should you keep before writing the disclosure?

Maintain a short internal AI-use record. It can include the tool and model/version (where identifiable); date and purpose; task instructions without personal information; output artifacts; changes made by the researcher; script versions; independent verification steps; and relevant institutional or journal approvals. When generative output varies by run, the record should document the actual analysis executed, not merely the prompt that suggested it (Sandve et al., 2013).

The disclosure and reproducibility record serve different purposes. The disclosure tells the reader where AI contributed. The record shows what the team actually did.

How should AI use be described in the Methods section?

If AI only suggested clearer prose, a Methods description might be unnecessary under a particular journal’s policy. If AI contributed to creating code, select variables, transform data or extract evidence, the work may need methodological detail even if an AI declaration appears elsewhere.

For statistical analyses, focus on the scientific decisions: outcome definition, study design, analysis population, statistical model, handling of missing values, uncertainty measures and prespecified versus exploratory decisions. Name the software and explain which parts of the workflow were verified by a human. The methods should be defensible without asking the reader to trust the tool.

Frequently asked questions

Disclosure itself does not establish that work is unacceptable. Acceptance depends on the journal’s policies and the quality and integrity of the research. Misrepresenting AI use can be a separate problem from whether the work is statistically sound (ICMJE, 2026).

No, not under ICMJE guidance. A tool cannot meet the authorship requirement to take responsibility for published claims (ICMJE, 2026).

Not necessarily. Follow your journal’s instructions and preserve a proportionate internal record. Never disclose private or restricted inputs merely to publish prompts.

A statistician can explain which analytical tasks were AI-assisted and how they were verified. Independent Statistical Review covers analyses prepared with AI assistance. The authors should confirm that every description is accurate and aligned with journal requirements.

Independent Statistical ReviewNeed a statistician to confirm the analysis before you submit? An in-house PhD-level statistician reviews the analysis behind your disclosure, so the Methods text describes what was actually done.

Request an Assessment

Independent Statistical Review

Independent Statistical Review is carried out in-house by PhD-level statisticians, who check method, assumptions, reproducibility and reporting. Authorship and responsibility for the study remain with the researcher. A statistically trained Research Coordinator reviews each request, and no commitment is required at the assessment stage.

References

  • International Committee of Medical Journal Editors. (2026). Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals: Use of AI by authors. https://www.icmje.org/recommendations/browse/artificial-intelligence/ai-use-by-authors.html
  • Sandve, G. K., Nekrutenko, A., Taylor, J., & Hovig, E. (2013). Ten simple rules for reproducible computational research. PLOS Computational Biology, 9(10), e1003285. https://doi.org/10.1371/journal.pcbi.1003285
  • Zielinski, C., Winker, M. A., Aggarwal, R., Ferris, L. E., Heinemann, M., Lapeña, J. F., Jr., Pai, S. A., Ing, E., Citrome, L., Alam, M., Voight, M., & Habibzadeh, F. (2024). Chatbots, generative AI, and scholarly manuscripts: WAME recommendations on chatbots and generative artificial intelligence in relation to scholarly publications. Current Medical Research and Opinion, 40(1), 11–13. https://doi.org/10.1080/03007995.2023.2286102

Last reviewed November 2026 by the OnlineSPSS Statistical Team. Worked examples use hypothetical values and are not client data.

NEXT STEP

Need this done for
your study?

Share your research in about a minute and receive a complimentary assessment the same day: scope, timeline and one fixed fee.

No Commitment Required

Strictly Confidential

Fixed Project Fee

In-House PhD-Level Statisticians

Same-Day Response

Serving Researchers Since 2014