As artificial intelligence (AI) tools become more common in academic writing, publishers, editors, and researchers worldwide have developed clear guidelines to ensure integrity, transparency, and accountability. Below are the main recommendations and ethical points that authors, reviewers, and editors should follow.
1. Before Submission: What Authors Must Do
Full Transparency About AI Use
Authors must clearly declare any use of generative AI tools—whether for idea generation, study design, data collection or analysis, writing, editing, or translation. This declaration should include the exact name of the tool, its version, and a specific description of how it was used. The statement should appear in a dedicated section, such as the "Methodology" or a separate "Declaration of AI Use." Hiding or under reporting AI use is considered a serious ethical breach and may lead to rejection or retraction.
AI Cannot Be Listed as an Author
According to the International Committee of Medical Journal Editors (ICME) and all major publishers, AI tools do not meet the criteria for authorship. They cannot take responsibility for the content, manage conflicts of interest, or agree to copyright terms. Therefore, listing an AI tool as a co‑author is strictly forbidden. AI may be acknowledged as a "research assistant" or "tool," but the final responsibility for every part of the manuscript lies entirely with the human authors.
Proper Citation of AI Outputs
If authors directly use text, ideas, images, or analyses generated by an AI, they must cite the output appropriately. The citation should include the tool’s name, version, date of use, and, if possible, a link to the interaction or prompt.
No Fabrication or Manipulation of Data
Using AI to create fake raw data or to alter actual results is absolutely prohibited. Generating references that do not exist or citing fabricated articles is a clear case of academic misconduct.
Human Review and Final Approval
Authors must carefully review and edit all AI‑generated content. AI outputs can be inaccurate, biased, or incomplete. Authors must confirm that the final manuscript, including any AI‑assisted text and images, contains no plagiarism and that all quoted material is properly referenced.
Originality and No Dual Submission
The manuscript must be the original work of the authors. Including unrelated individuals as co‑authors is not allowed. The manuscript should not have been published previously or be under consideration by another journal at the same time.
Correction Obligation
If an error is discovered at any stage—during peer review or after publication—the corresponding author must promptly notify the journal and cooperate to correct or retract the paper.
2. During Peer Review: Responsibilities of Reviewers
No Delegation of Final Judgment
Reviewers must not delegate their core evaluative tasks—critical assessment and final recommendations (accept, revise, or reject)—to AI. Scientific judgment requires deep understanding, contextual insight, and ethical reasoning that machines cannot provide.
Confidentiality and No Uploading Manuscripts
Submitted manuscripts are confidential. Uploading full texts, abstracts, or data into AI platforms for summarising or editing violates confidentiality, because these platforms may use the input data for training and could inadvertently disclose the work before publication.
Limited Use and Full Accountability
Reviewers may use AI only for limited, auxiliary purposes, such as improving the language of their own review report or clarifying technical concepts—without uploading manuscript content. In any case, the reviewer bears full responsibility for the accuracy and fairness of their report and cannot blame AI for any errors.
Inform the Editor
If reviewers use AI in any capacity, they should inform the editor. AI may help with summarising, but it cannot replace human critical thinking or decision‑making.
3. The Role of Editors and Journal Policies
Clear and Public Policies
Journals must publish explicit policies in their "Author Guidelines" that state what types of AI use are permitted and what are forbidden. These policies should be easy to find and understand.
Transparency in Editorial Work
Editors and editorial board members should also be transparent if they use AI tools during manuscript handling or editing.
Human‑Driven Decisions
Final decisions about acceptance or rejection, policy setting, and overall editorial judgment must remain in human hands. AI can support but never replace these decisions.
Monitoring and Follow‑Up
Editors should check the declared AI use during review. Content that appears to have undeclared AI assistance may be subject to plagiarism screening or may lead to rejection.
4. Consequences of Violations
Breaching AI policies can result in:
- Rejection of the manuscript, or retraction if already published.
- Notification to the author’s institution, funding body, or relevant authorities, following the Committee on Publication Ethics (COPE) guidelines.
- Serious action against cases of data fabrication, ghostwriting, or citation fraud, treating them as scientific misconduct.
5. Summary: A Practical Framework for Responsible Publishing
Based on the policies of major publishers (Elsevier, Springer, Wiley), COPE, ICMJE, and other international bodies, the following principles form a practical framework for responsible use of AI in scholarly publishing:
Transparency: Fully disclose all AI use at every stage of research and writing.
Human Responsibility: Human authors take full responsibility for the final content; AI cannot be an author.
Human Oversight: All AI outputs must be reviewed, edited, and approved by humans.
Originality and Integrity: AI should never replace the researcher’s own critical thinking, creativity, or scholarly analysis.
Confidentiality: Never upload an unpublished manuscript to a public AI platform.
Education and Awareness: Institutions and journals should provide training through workshops, webinars, and clear guides on both ethical and technical aspects of using AI in research.
دیدگاه خود را بنویسید