Generative Artificial Intelligence
Generative artificial intelligence is now part of ordinary academic work, and Peer Academics does not forbid it. What the journal requires is that its use is disclosed, that a human being remains accountable for every claim and every citation, and that no unpublished manuscript is ever entered into an external service. This policy covers large language models and chat assistants, agentic tools that search, browse, or run code on your behalf, AI writing and editing assistants, AI translation, AI image generation and editing, and AI tools that search or summarize literature. The version in force is the one published on this page on the date your manuscript was submitted.
For authors — disclosure
Record any use in the Use of generative artificial intelligence declaration described in Manuscript Structure, naming the tool and its provider, the version or the date of use, and what it was used for and in which part of the work. If you used none, say so in those words. Disclosure is published with the article.
- Disclose — drafting, rewriting, expanding, or summarizing any part of the text; translating your own text into English; summarizing or screening literature; generating or refactoring code used in the analysis; generating figures, illustrations, or synthetic data; extracting or coding data from sources.
- No disclosure needed — spelling and grammar checkers, reference managers, plagiarism or similarity tools, ordinary search engines, and statistical or analytical software used in the conventional way.
Where a tool did substantive analytical work, describe it in the Methods as you would any other instrument — name, version, dates, prompts or settings where they affect reproducibility — rather than only in the Declarations.
For authors — authorship and responsibility
An AI tool cannot be an author and cannot be credited as one, in any position; it cannot consent, approve a manuscript, or answer for it. See Authorship and Contributorship. You are fully responsible for everything you submit, including anything a tool produced. In particular:
- Every reference must be real, and you must have read the source. Language models invent plausible citations, quotations, page numbers, and identifiers. Verify each one in the original, not in the tool. A manuscript containing fabricated or unverifiable references is declined; if that is found after publication it is a retraction, not a correction.
- Every factual claim, number, and quotation is yours. "The tool produced it" is not a defense to error, misattribution, or plagiarism.
- Data may not be invented. Generated or simulated material may appear only where it is part of the research design and is described as such in the Methods; it may never be presented as observed data.
- Images and figures. Do not use generative tools to create or alter figures, photographs, or images that present evidence. Generative output is permitted only where it is itself the object of study or an explicit illustration, and must be labeled as such in the caption.
- Confidentiality and personal data. Do not enter participants' data, interview transcripts, clinical material, another researcher's unpublished work, or anyone else's confidential information into an AI service. The journal is governed by Romanian law and Regulation (EU) 2016/679 (GDPR); questions go to [email protected].
Undisclosed use that materially affected the manuscript is handled under Publication Ethics, as is any use that produced fabricated content.
For reviewers
- Do not enter a manuscript, or any part of one — the title, the abstract, a table, a figure, a paragraph — into any generative artificial intelligence service. A submission is confidential and unpublished; uploading it discloses it to a third party, may place it in a training corpus, and is a breach of confidentiality whatever the service's settings say. This applies to assistants built into browsers, email clients, note-taking apps, and code editors.
- Do not ask a tool to write, draft, or summarize your report. The editor is asking for your judgment, which cannot be delegated.
- You confirm compliance in the declarations part of the review form. If you used an AI tool in any way connected to the review, say so there and describe it; the editor decides whether the report can be used.
- A breach is treated as a confidentiality breach: the report is set aside, the reviewer is removed from the journal's reviewer pool, and the authors are told if their manuscript was exposed.
See also Reviewer Confidentiality.
For editors
- Editors are bound by the same prohibition: no manuscript, review report, or item of editorial correspondence is entered into an external AI service.
- Editorial decisions are taken by editors. No tool may generate a decision, a decision letter, or an assessment of a manuscript's merit.
- AI-detection output is not evidence. Detectors are unreliable, and they misclassify the writing of authors working in a second language particularly often. No manuscript is rejected, and no author is accused, on the basis of a detector score or because a text "reads like AI". Where an editor has a real concern, it is raised as a question to the authors about the disclosure and pursued on evidence such as fabricated references or claims unsupported by the cited sources.
- Editors may use ordinary administrative software; they may not use AI tools to select reviewers or to substitute for reading the manuscript.
Questions
If you are unsure whether a particular use must be disclosed, disclose it, or ask at [email protected] before you submit. Disclosing a use that turns out not to have required disclosure costs you nothing; failing to disclose one that did is misconduct.