Policy on the use of Artificial Intelligence and AI-Assisted Technologies

1. Purpose, scope, and principles

Romanian Journal of Public Affairs (“RJPA”) recognizes that artificial intelligence tools, including generative artificial intelligence systems, large language models (LLMs), image-generation systems, translation applications, AI-assisted analytical tools, and AI-assisted coding applications, may support research and scholarly writing.

The use of such tools does not remove the academic, ethical, professional, or legal responsibilities of authors, reviewers, and editors. This policy is guided by the principles of human accountability, transparency, research integrity, confidentiality, verifiability, respect for intellectual-property rights, and independent human scholarly judgment.

This policy applies to authors, reviewers, editors, editorial-board members, and all persons involved in RJPA’s editorial and publication processes.

2. Authorship and accountability

Artificial intelligence tools, including generative AI systems, must not be listed as authors or co-authors of a manuscript. AI tools cannot meet the requirements for authorship because they cannot approve the final version of a work, accept responsibility for its accuracy and integrity, disclose conflicts of interest, sign editorial declarations, or hold and manage copyright.

All named human authors remain fully responsible for:

  • the accuracy, originality, and coherence of the manuscript;
  • the validity of data, methods, analyses, code, and interpretations;
  • the verification of all references, citations, factual claims, and quoted material;
  • the absence of plagiarism, fabrication, falsification, or misrepresentation;
  • compliance with research-ethics requirements and applicable law;
  • compliance with copyright, licensing, and attribution requirements.

Authors may not rely on the use of an AI tool as an excuse for errors, unsupported claims, fabricated references, plagiarism, breaches of confidentiality, or any other form of publication misconduct.

3. Permitted uses and limitations

RJPA does not, as a general rule, prohibit the responsible use of AI tools, provided that such use complies with this policy and applicable legal and ethical requirements. AI use must be proportionate, transparent, and subject to meaningful human verification.

Permissible uses, subject to rigorous human review, include:

  • spelling, grammar, and punctuation correction;
  • improving readability, clarity, or stylistic consistency;
  • translation assistance, provided that the final text is fully checked by the authors or an appropriately qualified person;
  • code generation or code checking, provided that the code is tested, documented, and validated;
  • assistance with organizing materials, preliminary brainstorming, outlining an analysis plan, or drafting preliminary formulations;
  • documented methodological uses, including classification, coding, data processing, data analysis, and visualization, where such use is methodologically appropriate and sufficiently described to permit evaluation and, where applicable, replication.

The following uses are not permitted:

  • fabricating, inventing, altering, or unjustifiably completing empirical data;
  • including AI-generated references, citations, or sources that have not been verified by the authors in the original source;
  • presenting AI-generated content as empirical evidence, original analysis, or human scholarly judgment without clear disclosure;
  • using AI to fabricate interviews, survey responses, observations, documents, datasets, statistical results, case studies, or other empirical evidence;
  • using AI-generated or AI-manipulated images, maps, charts, tables, illustrations, or other materials in a manner that may mislead readers about the origin, authenticity, or meaning of the evidence;
  • entering confidential, unpublished, personally identifiable, sensitive, classified, or otherwise protected information into public or unauthorized AI systems.

4. Disclosure of AI use

Authors must disclose any substantial or relevant use of AI-assisted technologies in the production of a manuscript. The disclosure must enable editors, reviewers, and readers to understand whether and how an AI tool influenced the research process, analysis, writing, or presentation of findings.

Routine use of automated spelling, grammar, or punctuation correction normally does not require a separate disclosure, provided that the tool has not generated, substantially rewritten, fully translated, or materially altered the manuscript’s argument, interpretation, data, or conclusions.

In all other cases, authors must:

  • disclose the use of AI in the cover letter at submission;
  • describe the use of AI in the most relevant section of the manuscript;
  • include, where appropriate, a separate statement before the References section.

Where AI has been used for data collection, data cleaning, coding, classification, analysis, code generation or checking, tables, figures, or visualizations, the use must be described in the Methods section or an equivalent section. The description must contain sufficient detail to permit methodological assessment.

Where AI has been used for writing assistance, paraphrasing, translation, or substantial editing, the use must be disclosed in the Acknowledgments section or in a separate section entitled “AI and AI-Assisted Technologies Statement.”

The disclosure should include, as applicable:

  • the name of the tool;
  • the provider or developer;
  • the version, model, or date of use, where available;
  • the precise purpose of use;
  • the manuscript component or research stage affected;
  • the human verification, validation, and quality-control measures applied by the authors;
  • for methodological uses, the parameters, workflow, input data, and information necessary for assessment or reproduction, insofar as disclosure is compatible with ethical, legal, and contractual obligations.

Authors should retain reasonable records of relevant AI use, including the tool version, date of access, relevant instructions, input data, and outputs used, where necessary to verify the research process. RJPA may request this information during peer review or in response to concerns about research integrity.

Prompts or instructions supplied to AI systems do not need to be automatically published in the manuscript. They may be requested by the editorial office, included as supplementary material, or described methodologically where necessary for transparency, verification, or replication.

5. Data protection, confidentiality, intellectual property, and sources

Authors must review the terms of use, privacy policies, and data-retention or model-training practices of any AI tool used in relation to their research. Public or unauthorized AI services must not be used to process confidential or protected information.

Authors must not enter into public or unauthorized AI systems:

  • non-anonymized personal data;
  • information that could enable the re-identification of research participants;
  • sensitive personal data or other legally protected categories of information;
  • confidential information obtained from institutions, organizations, contractual partners, or research participants;
  • unpublished manuscripts, peer-review reports, editorial correspondence, or materials belonging to other authors.

Authors must directly verify every reference, citation, DOI, factual claim, attributed quotation, and source. AI-suggested references must not be included unless their existence, accuracy, relevance, and bibliographic details have been checked against the original source or an authoritative bibliographic record.

AI tools must not be cited as authors. When necessary for methodological transparency, an AI tool may be described in the AI disclosure statement, the Methods section, or an explanatory note, consistent with RJPA style requirements.

6. Images, figures, and visual materials

Any image, figure, map, diagram, illustration, or other visual material generated or substantially modified using AI must be explicitly disclosed. The disclosure must identify the tool, the nature of the intervention, and the authors’ role in checking the content.

AI must not be used to modify images, documents, screenshots, maps, or other empirical materials in a manner that changes the meaning of the evidence or may mislead readers.

For figures, charts, and tables derived from empirical data, authors must be able to provide, upon request and subject to ethical and legal constraints, the source data, code, processing steps, and explanations necessary to verify the reported findings.

7. Peer review and editorial use of AI

Submitted manuscripts, supplementary materials, peer-review reports, and editorial correspondence are confidential. Reviewers, editors, and editorial-board members must not upload, copy, submit, or transmit these materials, in whole or in part, to public, free, commercial, or unauthorized AI tools, including chatbots, large language models, translation services, or external analytical systems that may retain, reuse, or use data for model training.

An AI tool may be used in editorial or peer-review work only where:

  • the tool has been expressly approved by RJPA;
  • confidentiality, data security, and compliance with applicable law are ensured;
  • the manuscript, author data, and editorial information are not retained, reused, or used to train the model;
  • the use is compatible with the reviewer’s or editor’s obligations of impartiality and independence;
  • the responsible editor has been informed and the use has been appropriately documented.

Peer-review reports, editorial recommendations, and publication decisions must be based on the responsible person’s critical analysis, professional assessment, and independent judgment. Reviewers and editors must not delegate scientific assessment, originality assessment, methodological evaluation, editorial recommendations, or the complete drafting of a review report to an AI tool.

If a reviewer or editor uses an RJPA-approved AI tool for a limited technical or administrative function, that use must be disclosed to the Editor-in-Chief or editorial office, including the tool used and the purpose of use.

8. Detection, investigation, and editorial action

RJPA may use technical tools, editorial checks, expert assessment, and requests for clarification to investigate concerns relating to undisclosed or inappropriate AI use, plagiarism, data fabrication, or other breaches of research integrity.

The output of an AI-content detector does not, by itself, constitute evidence of misconduct and will not be the sole basis for an editorial decision. Any editorial decision will be based on documented, contextual, human assessment.

Where relevant AI use has not been disclosed, where inaccurate information has been supplied, or where confidentiality, citation, data-integrity, or publication-ethics standards have been breached, RJPA may request explanations and supporting documentation, require revision, suspend review, reject the manuscript, or, after publication, take appropriate editorial action, including correction, expression of concern, or retraction.