Since its introduction, artificial intelligence (AI) has rapidly become one of the most disruptive technologies in history. Perhaps nowhere is this truer than with scientific publishing. Although medical science has always valued novelty, debate, and innovation, AI presents unique challenges which risk undermining scientific integrity and faith in trusted institutions.
Beyond its role as a helpful resource and fact checker, AI is increasingly the primary source of information for many individuals. And despite its novelty, it is already perceived as being more definitive than historical gold standard resources such as clinicians, researchers, and institutions. Perhaps even more importantly, as researchers rely more and more on AI tools, they may lose the ability to perform analyses themselves, thus losing the ability to challenge AI responses directly.
Arguably one of the most notable concerns with AI and publishing relates to data falsification. Currently, a poorly written paper with limited methodological details and inconsistent results is quickly rejected. However, with AI, an average middle-schooler can create falsified scientific publications that would fool even the most astute editorial teams. This developing challenge will most certainly necessitate modification of existing review processes, which may increasingly include evaluating the authorship teams themselves as well as other novel validation measures.
In contrast to the above concerns, AI offers several advantages to the scientific community. It empowers authors to better assimilate literature, optimize study design, identify scientific gaps, perform analyses, and communicate results. It also eliminates geographic/language barriers, which have historically limited scientific dissemination. Although some might suggest that AI-generated content jeopardizes the scientific process, it’s helpful to recognize that the goal of publication is to accurately communicate methods and results and not necessarily to test one’s ability at creative writing.
AI tools will also be increasingly integrated in the review process. In this manner, they can help reviewers better identify plagiarized material, inaccuracies, missing/incorrect references, and other “red flags.” Although careful consideration must be given to copyrights and other protections, these innovations could readily be compared to how CT scans advanced detection of abnormalities over the less detailed X-ray films alone.
Regardless of how one feels about AI with scientific publishing, it is here to stay. And as with other historical innovations, the wisest approach will likely be to proactively direct its integration in responsible ways rather than to constantly resist inevitable progress.
Ethical use of AI for authors
The ethical use of AI in publishing requires clear boundaries between acceptable supportive functions and inappropriate practices. Authors may use AI tools for specific purposes, such as language editing, grammar correction, or improving clarity of expression and presentation of data, as long as the original intellectual content, interpretation, and conclusions are produced by humans. In this context, AI can be a helpful tool, especially when writing in a non-native language, to better communicate scientific ideas. Conversely, the use of AI for generating, manipulating, or fabricating data, results, or citations represents scientific misconduct, undermines research integrity, and cannot be considered an ethical use.1
Regarding authorship and accountability, AI systems do not meet the criteria for authorship, as they lack agency, responsibility, and the ability to provide consent or take legal accountability. Therefore, as also reported by the International Committee of Medical Journal Editors (ICMJE; https://www.icmje.org/recommendations), AI cannot be listed as an author or co-author. All human authors remain fully responsible for the accuracy, validity, and originality of the work, including any content that has been generated or supported by AI tools. This also includes verifying outputs, checking for possible bias or hallucinated information, and ensuring compliance with ethical and legal standards.
Transparent disclosure of AI use is essential to maintain trust and scientific rigor. Authors should clearly describe the nature and extent of AI assistance, including which tool was used and for what purpose. Such information should be included in a clearly identifiable section of the manuscript. The ISSM has developed guidelines for how to use and document the use of AI when publishing in our journals (Artificial Intelligence AI Policy for Authors and Peer Reviewers). This supports editorial oversight and helps ensure that metadata are captured correctly for indexing.
Ethical use and confidentiality for reviewers
The possible application of AI in the peer-review process presents several challenging issues to be discussed. AI tools can help potential reviewers and accelerate several aspects of the process, including language correction, checking for plagiarism, and data manipulation. In addition, AI allows for rapid assessment of manuscript adherence to publisher and journal checklists. Furthermore, AI can verify data, investigate protocol adherence, and discover possible methodological problems. Finally, AI can compare the consistency of data and terminology between abstract and manuscript.2 All these feasible advantages can save time, allowing reviewers to focus their analysis on evaluating the scientific content of the paper.
Despite the aforementioned advantages, the use of AI in the peer review process entails several other debatable and ethical aspects. First, uploading unpublished research data on AI systems can create crucial ethical risks related to the security of the data in terms of privacy and intellectual property protection. However, the most important concern lies in the ability of AI tools to critically assess complex scientific content. Similarly, there is still conflict concerning the capacity of AI to recognize the novelty or clinical importance of research. Accordingly, the latter issues can depend on several aspects, including clinical experience and patient and social needs, which can vary according to topic and culture.2
In addition, the possibility should be recognized that depending on their initial training, AI tools could present crucial biases related to specific clinical conditions or methodologies. This point further emphasizes other critical points. Accordingly, introducing AI-derived information or recommendations that are misleading or not completely correct can generate ethical and legal concerns related to the source primarily responsible for the bias. The latter possibility eventually results in an increase rather than a decrease in workload because of the need to verify AI suggestions.
Hence, at the present time, AI improves several aspects of the peer review process, supporting reviewers in generating templates and simplifying several other general aspects. However, the real role of available AI tools in assessing the clinical and experimental significance of an overall manuscript remains insufficiently verified and the object of intense debate.
Summary
Artificial intelligence is a transformative change that is affecting the scientific publishing world and that introduces several potential opportunities and challenges. The tools offer the possibility of improving study designs, efficiency, and execution and may help better communicate findings, particularly when published in a non-native language. These new tools may also expedite the overall review process and help identify areas needing further investigation or correction. However, AI also introduces a significantly greater risk of publishing falsified or incorrect data. Additionally, it raises new potential issues with confidentiality, intellectual property protection, and bias. In this context, it is important to recognize that new frameworks will need to be established to better outline the appropriate rules and uses of AI in the scientific publishing space. Among other things, such a framework should include an underlying principle of the ethical use of AI, where human authors maintain responsibility for content, accuracy, interpretations, and conclusions. Through the implementation of these concepts, AI may serve as a supportive and facilitative tool while maintaining the academic integrity and rigor that define the scientific process.
Contributor Information
Annamaria Giraldi, Sexology Clinic, Mental Health Center Copenhagen, Bispebjerg & Frederiksberg Hospital, Copenhagen 2400, Denmark; Department of Clinical Medicine, University of Copenhagen, Copenhagen 2200, Denmark.
Landon Trost, Male Fertility and Peyronie’s Clinic, Orem, UT 84057, United States; Cell Biology and Physiology, Brigham Young University, Provo, UT 84602, United States.
Giovanni Corona, Endocrinology and Metabolic Disease Unit Romagna, Forlì-Cesena 47121, Italy.
Funding
None declared.
Conflicts of interest
None declared.
References
- 1. Kwon D. Is it OK for AI to write science papers? Nature survey shows researchers are split. Nature. 2025;641(8063):574–578. 10.1038/d41586-025-01463-8 [DOI] [PubMed] [Google Scholar]
- 2. Perlis RH, Christakis DA, Bressler NM, et al. Artificial intelligence in peer review. JAMA. 2025;334(17):1520. 10.1001/jama.2025.15827 [DOI] [PubMed] [Google Scholar]