Editorial policies

Policy on the use of artificial intelligence tools

Contents

• 1. Introduction
• 2. Guiding principles for the ethical use of AI
• 3. Guidelines for participants in the editorial process
• 4. How to disclose the use of AI
• 5. Minimum information required in the methodology
• 6. Effective date and review
• References

1. Introduction

This policy governs the use of artificial intelligence (AI) tools, including large language models (LLMs), chatbots and image generators, at every stage of the editorial process of Población y Salud en Mesoamérica (PSM), a biannual, continuous-publication electronic academic journal of the Centro Centroamericano de Población (CCP), published in collaboration with the Instituto de Investigaciones en Salud (INISA), with the University of Costa Rica as publisher.

The policy applies to all types of work the journal receives (scientific article, theoretical article, technical report, literature review and essay) and to everyone involved in the editorial process: authors, reviewers, editors and members of the Editorial Board. It complements the journal's Ethics and Good Editorial Practices Policy and is grounded in the guidance of the Committee on Publication Ethics (COPE), the International Committee of Medical Journal Editors (ICMJE) and the Heredia Declaration.

Its purpose is to promote the quality and clarity of manuscripts without replacing critical analysis or rigorous research. Because the journal disseminates population research from a demographic perspective and public health research, which frequently relies on microdata from censuses, surveys, administrative records and health data, this policy pays particular attention to protecting the personal and sensitive data of research participants.

2. Guiding principles for the ethical use of AI

The use of AI in Población y Salud en Mesoamérica is based on the following principles, consistent with the principles of transparency, quality, and recognition and authorship set out in the journal's Ethics Policy:

• Transparency and traceability:Any use of AI that contributes to the preparation of the manuscript (text, literature review, analysis, translation, images, tables or programming code) must be disclosed explicitly, clearly, sufficiently and verifiably. Transparency builds trust in the scientific process and allows other researchers to replicate or evaluate the work effectively.
• Truthfulness and integrity of the scientific record:The use of AI to fabricate, falsify, alter or manipulate data, results, evidence, citations, references or images in a misleading way is absolutely rejected. Such conduct constitutes scientific fraud under the journal's Ethics Policy.
• Legal and human responsibility:Intellectual, ethical, academic and legal responsibility for the submitted and published content rests exclusively with the authors. AI cannot assume legal responsibility and has no legal personality; it must therefore always be used under the direct supervision and responsibility of the authors.
• Confidentiality and protection of content:Material submitted to the journal and the peer review process are strictly confidential. Sharing manuscripts or parts of them with AI tools that compromise confidentiality is prohibited, especially through free services or services without institutional data protection guarantees.
• Protection of personal and sensitive data:It is prohibited to enter into AI tools any microdata, databases, interview transcripts or other information that could directly or indirectly identify research participants, as well as health data or other sensitive data, unless such data have been properly anonymized and authorization has been obtained from the relevant research ethics committee and the data custodian, in accordance with Costa Rica's Law No. 8968 on the Protection of Individuals with regard to the Processing of their Personal Data and the regulations of the University of Costa Rica.
• Prohibition of manipulation through hidden prompts:¹ Inserting hidden instructions or any mechanism into manuscript files intended to manipulate editorial judgment or reviewers' assessments is strictly prohibited. If confirmed, the manuscript will be rejected.

3. Guidelines for participants in the editorial process

A) Responsibilities of authors

• AI cannot be an author:The journal adopts the four ICMJE authorship criteria (substantial contribution, drafting or critical revision, approval of the final version, and accountability for all aspects of the work). AI systems cannot meet these criteria, particularly accountability, and therefore cannot be listed as authors or co-authors of an article.
• AI cannot be cited as a reference:AI may not appear in the reference list of any article. Sources such as "ChatGPT/AI as a source of knowledge" are not accepted; authors must cite only verifiable academic sources. Every reference must be checked by the authors, since these tools can generate non-existent citations.
• Human review and correction:Authors are fully responsible for reviewing, verifying and correcting any AI-generated output, actively checking for biases, false data or nonsensical information that these tools may produce.
• Language and copy editing:Using AI to improve the readability, grammar, spelling, punctuation and tone of text written by the authors does not need to be disclosed, provided that it involves no generative editorial work or autonomous content creation. Full or partial translation of the manuscript (including the abstract) using AI must be disclosed.
• Use in data, analysis and images:Generative AI may not be used to generate, fabricate or fill in scientific data, or to create or alter graphs, figures, maps or results tables. This prohibition does not extend to established machine learning techniques and statistical methods (for example, predictive models, classification methods or natural language processing for coding variables), which are legitimate methodological tools and must be described in the methods section. Other exceptions will be accepted only when methodologically justified, such as research on AI itself or the use of synthetic data for statistical disclosure control, and they must be described in detail in the methods section.
• Programming code:If AI was used to write, debug or optimize analysis code (R, Stata, Python, SPSS or other), authors must verify that the code works and that the results are valid, and must disclose this use. The journal recommends making the code available as supplementary material or in an open repository.
• Automated generation of articles:Using AI tools to automatically generate complete articles or complete sections of articles is strictly prohibited.
• Mandatory disclosure:Everyone submitting a manuscript from [effective date] onward must include an explicit statement on the use or non-use of AI, using the form the journal makes available. This statement is incorporated into the manuscript as indicated in section 4 and is published with the article.

B) Responsibilities of reviewers

• No delegation of the review:Reviewers may not delegate the critical evaluation or the writing of the review report to AI. Evaluation is a human, expert, personal and non-transferable responsibility.
• Prohibition on uploading manuscripts:In line with the commitment to absolute confidentiality established in the journal's Ethics Policy, reviewers are strictly prohibited from uploading the manuscripts under review, tables, figures, unpublished data or supplementary material to generative AI tools (especially free or open tools), given the high risk of breaching the confidentiality and intellectual property of the authors. This includes AI-generated text detectors.
• Auxiliary use and disclosure:AI may be used solely as an auxiliary aid to correct the style or wording of the reviewer's own report. If used for this purpose, the reviewer must disclose it transparently, specifying the tool and confirming that no confidential material was exposed.
• Suspected undisclosed use:If a reviewer suspects that the manuscript contains undisclosed AI-generated content or hidden prompts, they must report it to the editorial team in the review form, without carrying out their own checks using external tools.
• Inclusion in the review form:The guidelines in this section and the reviewer's AI use statement are incorporated into the journal's peer review form, so that all reviewers are aware of and accept them when conducting the review, without depending on whether they consult the policies published on the website.

C) Guidelines for the editorial team

• Oversight and control:Editors will oversee compliance with this policy by authors and reviewers. In case of doubt or ambiguity, they may request additional clarifications, logs or records of prompts before accepting the manuscript.
• Human decision-making:All editorial decisions (acceptance, rejection, requests for revisions, assignment of reviewers) will be made exclusively by members of the editorial team and the Editorial Board, and never by automated systems.
• Internal institutional use:The editorial team may use institutionally licensed AI tools (for example, Microsoft 365 Copilot) or the similarity checking tool the journal already uses (Turnitin) for administrative tasks, metadata management or plagiarism detection, under strict safeguards of confidentiality and data minimization, in accordance with current institutional regulations. Results from AI-generated text detectors will not, on their own, be considered sufficient proof of non-compliance.
• Journal statement:The journal will publish on its website the list of AI tools used by the editorial team in manuscript management and the purpose of their use, and will make the disclosure form available to authors together with the author guidelines.
• Non-compliance:Failure to provide the statement or deceptive use of AI will be considered a serious breach of academic integrity and will be handled in accordance with the journal's Ethics Policy. If the article has not been published, it may be rejected; if it has already been published, it may be corrected or retracted. If bad faith is proven, the disciplinary procedure of the University of Costa Rica will be followed or, where the institutional affiliation is different, the authorities of the respective institution will be notified.

4. How to disclose the use of AI

Placement in the manuscript

1. At the beginning of the manuscript: The statement will be placed after the abstract (in Spanish and English) and the keywords, and before the introduction, under the heading "AI Use Statement". It will indicate whether AI was used and, if so, the tool used, its purpose and the sections of the manuscript in which it was used.
2. In the methodology section: When AI forms part of the methodological design, systematic review, data processing, coding, modeling or analysis, its use will be described in greater detail in the methodology section, with the minimum information indicated in section 5.

Examples of correct wording:

Example 1: Manuscript that did NOT use artificial intelligence

I declare that no artificial intelligence was used during the research process or the preparation of this manuscript. As author(s), the intellectual, ethical and academic responsibility for the submitted and published content rests exclusively with me/us.

Example 2: Manuscript with substantive or supporting use in the research

We declare that the tool ChatGPT (OpenAI, GPT-4o model, May 2026 version, used between 10 and 15 May 2026) was used in the preparation of this manuscript to debug the R code used to analyze the survey microdata (methodology section) and to support the translation of the abstract into English. The microdata and any information that could identify participants were not uploaded to the tool. Use was strictly limited to the tasks described. As authors, we fully verified, cross-checked and edited the output generated by the tool, confirmed that the code produced correct results, and take full responsibility for the final content of the manuscript.

Example 3: Statement required from reviewers (in their report)

To support the writing and improve the clarity of this review report, I used Microsoft 365 Copilot (institutional version) solely for grammatical correction of my comments. The manuscript, its data and any confidential information about the article under review were not uploaded to the tool. The critical analysis, substantive evaluation and conclusions of the report are entirely my responsibility as a reviewer.

5. Minimum information required in the methodology

If AI is involved in processing the research, the manuscript must rigorously detail the following in the methods section:

1. The exact name of the tool and the institutional or commercial provider.
2. The model, specific version and dates on which the system was used.
3. The purpose and scope of use, clearly distinguishing between support on form (style) and on substance (data analysis, coding, programming, etc.).
4. The inputs provided (general prompts or anonymized data) and the confidentiality safeguards applied.
5. Data handling: confirmation that no identifiable personal data were entered and, where applicable, reference to approval by the research ethics committee and authorization from the data custodian.
6. The human oversight and verification strategy used to detect biases, hallucinations or accuracy errors in the tool.

Example of a description in the methodology section:

For the preliminary classification of open-ended responses on reasons for not using health services (n = 1,248), Microsoft 365 Copilot was used (institutional license of the University of Costa Rica, accessed between 3 and 7 March 2026). Before use, the responses were anonymized by removing names, places and any other identifiable data, in accordance with the protocol approved by the research ethics committee. The tool was provided with the study codebook and asked to assign each response to one of its eight categories. Two researchers independently reviewed all assignments; agreement between the tool's classification and the human review was κ = 0.81, and discrepancies were resolved by consensus. The final classification corresponds to the human review. The prompts used are included as supplementary material.

6. Effective date and review

This policy takes effect on [effective date] and will be reviewed periodically by the editorial team of Población y Salud en Mesoamérica, in light of developments in AI tools and international recommendations on ethics in scientific publishing. Cases not covered will be resolved in accordance with the journal's Ethics Policy and COPE guidance.

References

Asamblea Legislativa de la República de Costa Rica. (2011). Ley de Protección de la Persona frente al Tratamiento de sus Datos Personales, N.° 8968 [Law on the Protection of Individuals with regard to the Processing of their Personal Data, No. 8968].

Committee on Publication Ethics. (2023). Authorship and AI tools: COPE position statement. https://publicationethics.org/cope-position-statements/ai-author

International Committee of Medical Journal Editors. (2024). Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals. https://www.icmje.org/icmje-recommendations.pdf

Penabad-Camacho, L., Penabad-Camacho, M. A., Mora-Campos, A., Cerdas-Vega, G., Morales-López, Y., Ulate-Segura, M., Méndez-Solano, A., Nova-Bustos, N., Vega-Solano, M. F., & Castro-Solano, M. M. (2024). Heredia Declaration: Principles on the use of Artificial Intelligence in Scientific Publishing. Revista Electrónica Educare, 28(S), 1-10. https://doi.org/10.15359/ree.28-S.19967

Población y Salud en Mesoamérica. (n.d.). Ethics and good editorial practices policy. University of Costa Rica. https://revistas.ucr.ac.cr/index.php/rpsm/ethicpolicy

Revista ALP. (n.d.). Política de uso de inteligencia artificial y Grandes Modelos Lingüísticos (LLM) [Policy on the use of artificial intelligence and large language models (LLMs)]. University of Costa Rica.

Revista Tecnología en Marcha. (2024). Política editorial sobre el uso responsable de Inteligencia Artificial (IA) en la revista Tecnología en Marcha [Editorial policy on the responsible use of artificial intelligence (AI) in the journal Tecnología en Marcha]. Editorial Tecnológica de Costa Rica, Instituto Tecnológico de Costa Rica. https://revistas.tec.ac.cr/index.php/tec_marcha/libraryFiles/downloadPublic/115

Springer Nature. (2023). Artificial intelligence (AI) in research and publishing. https://www.springer.com/de/editorial-policies/artificial-intelligence--ai-/25428500

Uniandes. (n.d.). Política de uso de herramientas de inteligencia artificial en las revistas científicas de UNIANDES [Policy on the use of artificial intelligence tools in UNIANDES scientific journals].

Contact: revista.ccp@ucr.ac.cr